Identification of Genes Periodically Expressed in the Human Cell

Molecular Biology of the CellVol. 13, 1977–2000, June 2002

Identification of Genes Periodically Expressed in theHuman Cell Cycle and Their Expression in Tumors□D

Michael L. Whitfield,* Gavin Sherlock,* Alok J. Saldanha,* John I. Murray,*Catherine A. Ball,* Karen E. Alexander,† John C. Matese,*Charles M. Perou,‡ Myra M. Hurt,† Patrick O. Brown,§�¶ andDavid Botstein*¶

*Departments of Genetics and §Biochemistry, �Howard Hughes Medical Institute, Stanford UniversitySchool of Medicine, Stanford, California 94305; †Department of Biomedical Sciences, College ofMedicine, Florida State University, Tallahassee, Florida 32306; and ‡Department of Genetics,Lineberger Comprehensive Cancer Center, University of North Carolina at Chapel Hill, Chapel Hill,North Carolina 27599

Submitted January 8, 2002; Revised February 26, 2002; Accepted March 7, 2002Monitoring Editor: Mark J. Solomon

The genome-wide program of gene expression during the cell division cycle in a human cancercell line (HeLa) was characterized using cDNA microarrays. Transcripts of �850 genes showedperiodic variation during the cell cycle. Hierarchical clustering of the expression patterns revealedcoexpressed groups of previously well-characterized genes involved in essential cell cycle pro-cesses such as DNA replication, chromosome segregation, and cell adhesion along with genes ofuncharacterized function. Most of the genes whose expression had previously been reported tocorrelate with the proliferative state of tumors were found herein also to be periodically expressedduring the HeLa cell cycle. However, some of the genes periodically expressed in the HeLa cellcycle do not have a consistent correlation with tumor proliferation. Cell cycle-regulated transcriptsof genes involved in fundamental processes such as DNA replication and chromosome segrega-tion seem to be more highly expressed in proliferative tumors simply because they contain morecycling cells. The data in this report provide a comprehensive catalog of cell cycle regulated genesthat can serve as a starting point for functional discovery. The full dataset is available athttp://genome-www.stanford.edu/Human-CellCycle/HeLa/.

INTRODUCTION

Proper regulation of the cell division cycle is crucial to thegrowth and development of all organisms; understandingthis regulation is central to the study of many diseases, mostnotably cancer. Many of the diverse molecular processesneeded to duplicate a cell, such as DNA replication, moni-toring of replication fidelity, and accurate segregation ofchromosomes to daughter cells, are characteristically aber-rant in cancer cells. These processes and their regulationhave been extensively investigated at the molecular level(reviewed in Stillman, 1996; Nurse, 2000; Shah and Cleve-land, 2000; Hinchcliffe and Sluder, 2001; Wittmann et al.,

2001). Characterization of the genome-wide transcriptionalprogram of the cell division cycle in mammalian cells is acritical step toward understanding the basic cell cycle pro-cesses and their role in cancer.

The genome-wide transcriptional program during the cellcycle has been investigated in a wide range of organisms,including budding yeast (Cho et al., 1998; Spellman et al.,1998), bacteria (Laub et al., 2000), primary human fibroblasts(Iyer et al., 1999; Cho et al., 2001), mouse fibroblasts (Ishida etal., 2001), and human HeLa cells (Crawford and Piwnica-Worms, 2001). In synchronized cultures of the budding yeastSaccharomyces cerevisiae, 800 genes were found to be period-ically expressed (Spellman et al., 1998) and in synchronizedcultures of the bacterium Caulobacter crescentus, 553 period-ically expressed genes were identified (Laub et al., 2000). Thehuman cell cycle was previously studied first by observingthe genes whose expression changes when primary fibro-blast cultures were stimulated with serum to reenter the cellcycle from a serum-deprived, resting state. Interpretation ofthese studies is complicated by the subsequent realization

□D A complete data set for this article is available at www.molbiol-cell.org. Article published online ahead of print. Mol. Biol. Cell10.1091/mbc.02–02–0030. Article and publication date are atwww.molbiolcell.org/cgi/doi/10.1091/mbc.02–02–0030.

¶ Corresponding authors. E-mail addresses: [email protected] or [email protected].

© 2002 by The American Society for Cell Biology 1977

that the major gene expression program observed is morethan a simple cell cycle response, but rather is combinedwith a prominent wound-healing response (Iyer et al., 1999).By modifying this approach, Ishida et al. (2001) identified578 genes that were induced upon serum stimulation andprovided evidence of cell cycle regulation based on thepattern of expression after release from a cell cycle arrestwith hydroxyurea, which blocks DNA replication. A studyof the G2 DNA damage checkpoint in HeLa cells synchro-nized by a double thymidine block identified transcripts thatshowed differential expression from S to G2/M phases butfailed to identify a specific transcriptional response associ-ated with the G2 DNA damage checkpoint (Crawford andPiwnica-Worms, 2001). Finally, �700 genes were identifiedas cell cycle regulated in primary human fibroblasts syn-chronized by a double thymidine block, which attempts toavoid the confounding serum stimulation response alto-gether (Cho et al., 2001).

Herein, we report the extension of these results by syn-chronizing HeLa cells not only by double thymidine blockbut also by two additional methods, a thymidine-nocoda-zole block and a physical method, mitotic shake-off. Bycombining data from experiments using these three differentsynchronization methods in five independent experiments,we identified �850 genes that are periodically expressedduring the cell cycle. Using this list of genes we were able toshow that most of the genes previously associated with theproliferative state of tumors are among the genes we iden-tified as periodically synthesized during the human celldivision cycle.

MATERIALS AND METHODS

Cell Culture, Synchronization, and RNA PreparationHeLa S3 cells were plated (2 � 106 cells) in 150-mm tissue culturedishes in DMEM with 10% fetal bovine serum and 100 U of peni-cillin-streptomycin (Invitrogen, Carlsbad, CA). Cells were arrestedin S phase by using a double thymidine block or in mitosis with athymidine-nocodazole block essentially as described previously(Whitfield et al., 2000) and in the supplemental material. Poly(A)RNA was prepared from cells collected at intervals (typically 1–2 h)by lysing cells directly on the plate with the FastTrack 2.0 mRNAisolation kit (Invitrogen). Synchrony was monitored by flow cytom-etry analysis of propidium iodide-stained cells (Stanford SharedFACS Facility, Stanford University School of Medicine, Stanford,CA).

Mitotic cells were collected every 10 min with an automated cellshaker (Eliassen et al., 1998), stored on ice, and plated at 2-h intervalsin fresh prewarmed media (at least 106 cells for each time point).Because only a small number of cells can be obtained by mitoticshake-off, total RNA was prepared using ULTRASPEC RNA isola-tion system (Biotecx, Houston, TX). The number of cells in S phaseat each point was determined using a 5-Bromo-2�-deoxyuridine(BrdU) Labeling and Detection Kit I (Roche Applied Science, Indi-anapolis, IN).

Reference RNA was prepared from asynchronously growingHeLa cells using TRIzol (Invitrogen) and poly(A) RNA isolated byaffinity chromatography on oligo-dT cellulose (Amersham Bio-sciences, Piscataway, NJ). Poly(A) RNA was used as a reference inall experiments except the mitotic shake-off, where total RNA waslabeled.

cDNA Synthesis and Microarray HybridizationRNA from synchronous cells was reverse transcribed into Cy5-dUTP (Amersham Biosciences)-labeled cDNA, and reference RNA

reverse transcribed into Cy3-dUTP (Amersham Biosciences)-labeledcDNA by standard methods (Eisen and Brown, 1999) (Details areavailable at http://genome-www.stanford.edu/Human-CellCycle/HeLa/.) Total RNA samples from cells collected in the mitoticshake-off experiment, and total reference RNA were first amplifiedusing a modified Eberwine protocol before cDNA synthesis (Wanget al., 2000) then labeled cDNA was prepared from the amplifiedRNA.

Spotted cDNA microarrays, containing 22,692 elements repre-senting �16,332 different human genes or containing 43,198 ele-ments representing �29,621 genes (estimated by UNIGENE clus-ters), were manufactured in the Stanford Microarray Facility(http://www.microarray.org). Equal amounts of Cy5- and Cy3-labeled cDNA were hybridized to spotted cDNA microarrays andscanned using a GenePix 4000A Scanner (Axon Instruments, UnionCity, CA). Detailed protocols are available at http://brownlab.stanford.edu/protocols.html/.

Data ProcessingData were extracted by superimposing a grid over each array usingGenePix 3.0 software (Axon Instruments). Spots of poor quality,determined by visual inspection, were removed from further anal-ysis. Data collected for each array were stored in the StanfordMicroarray Database (SMD) and are available from SMD at http://genome-www.stanford.edu/microarray/ (Sherlock et al., 2001).

Only features with signal intensity at least 20% above backgroundin both Cy5 and Cy3 channels and for which adequate quality datawere obtained for at least 80% of the samples in a given time coursewere analyzed further. Data points that did not meet these criteriaare blank in the primary data tables. Log2 (Cy5/Cy3) was retrievedfor each data point and used for all analysis, where (Cy5/Cy3) is thenormalized ratio of the background-corrected intensities, as definedin SMD (Sherlock et al., 2001).

Because of systematic differences between experiments (e.g., ar-ray batch, labeling methods, and synchronization methods) eachtime course was centered independently by filtering out the first,most significant eigengene (Alter et al., 2000), which was a domi-nant, constant vector. Because singular valve decomposition (SVD)requires a full data matrix, missing data points were estimatedusing a k-nearest neighbors algorithm (Troyanskaya et al., 2001)with k � 12. These imputed values were used throughout theanalysis and but were restored to “unknown” status in the figuresand left blank in the primary data tables (http://genome-www.stanford.edu/Human-CellCycle/HeLa/).

Identification of Periodically Expressed TranscriptsA Fourier Transform (eqs. 1–3) was applied to the data for eachclone in an experiment (Spellman et al., 1998), and the resultingvector (C, eq. 3) of the sine (A) and cosine (B) coefficients was stored,where T is the cell cycle period, t is the time after release, � is thephase offset, and ratio(t) is the normalized Cy5/Cy3 expression ratioat time t. The value of � was initially set to 0. The values obtainedfor C were determined over a range of 40 values of T equally spaced1 h above and below the estimated cell cycle period and the result-ing values averaged and stored.

A � � sin�2��t/T� � ��log2�ratio�t�� (1)

B � � cos�2��t/T� � ��log2�ratio�t�� (2)

C � �A, B (3)

D � �A2 � B2 (4)

The optimal cell cycle period was determined by finding the valueof T where the largest numbers of genes pass an arbitrary magni-tude cutoff (D, eq. 4). Fourier transforms were applied to the dataseries for each gene (eqs. 1–3), with equally spaced values of T, from

M.L. Whitfield et al.

Molecular Biology of the Cell1978

0 to 40 h in 15-min increments. The number of genes whose mag-nitude (D, eq. 4) exceeded the arbitrary cutoff of 3, 5, or 7 wasplotted and a period (T) was chosen that maximized the numbers ofgenes exceeding our arbitrary thresholds. In most cases, the deter-mined value of T was consistent with the data obtained by flowcytometry.

Because each experiment does not start at exactly the same pointin the cell cycle, an offset (�, eqs. 1 and 2) was calculated for eachdataset relative to the first double thymidine arrest. The magnitudesof the Fourier transform (D, eq. 4) for the 1000 highest scoring cloneswere summed using different values of �, equally spaced between 0and 2�. The offset that gave the highest average combined magni-tude (D, eq. 4) between the two datasets for these 1000 genes wasthen used. The Fourier transform was then repeated on the remain-ing datasets with the following values of and �: Thy-Thy 2 (T �15.5 h, � � 0.5 rad), Thy-Thy 3 (T � 15.4, � � 0.0 rad), Thy-Noc (T �18.5, � � 3.2 rad), and mitotic selection (T � 24.5, � � 3.5 rad). Thevectors C (eq. 3) for all five datasets were then summed and thegenes ranked according to the magnitude (D, eq. 4) of their combinedvectors. Note, the Thy-Noc and mitotic shake-off experiments, whicharrest cells in mitosis, have offsets of approximately half a cycle (�radians) from Thy-Thy 1, which arrest cells at G1/S.

Because the gene expression profiles of many cell cycle genes do notprecisely match sine and cosine curves, the expression profile of eachgene was correlated to an idealized vector obtained from known genesexpressed in each cell cycle phase (G1/S, S, G2, and G2/M) as definedin Figure 2A. Using a standard Pearson correlation, each gene receiveda peak correlation score defined as the highest absolute value correla-tion between one of the four idealized vectors and its expression profile(Spellman et al., 1998). The absolute value of the peak correlation wasused to scale the magnitude of the vector (C, eq. 2) generating a“periodicity score” for each gene (Table 1).

To estimate the minimum periodicity score for a cell cycle regu-lated gene, the analysis described above was repeated on random-ized data. The data were randomized either within rows only, orwithin both rows and columns, for each of the five datasets starting

with the imputed, SVD-centered data. The Fourier transform andcorrelations were applied using the previously calculated values ofT and �; the resulting vectors (C, eq. 3) were combined for eachdataset. The magnitude (D, eq. 4) of the Fourier transform wasscaled by each “gene’s” peak correlation to one of the four idealexpression profiles. This analysis, including the data randomization,was repeated 10 times and the scores combined by averaging thescore for each of the highest scoring “genes” from each randomiza-tion, followed by the second highest, third highest, etc. The esti-mated false positive rate at a given periodicity score is the numberof genes that obtain at least that score in the randomized data. Wechose a minimum periodicity score of 3.29, which gave us 1333clones at an initial false positive estimate of 1% when the data wererandomized in rows. Repeating this analysis 10 times gave anestimated 10 false positives (0.75%; periodicity scores of 5.18–3.33)when the data were randomized only within rows, and two falsepositives (0.15%; periodicity scores of 3.72 and 3.30) when the datawere randomized in both rows and columns.

The false positive estimate, calculated above, is likely to under-estimate the true false positive rate because it does not take intoaccount genes that received a high Fourier score because theyexhibited a sinusoidal pattern in only part of a time course. To filterout genes that did not show periodic expression, autocorrelationsfor each 1333 clones were calculated (eq. 5). The autocorrelation A isequal to the summation over all times t of the product of the ratio att multiplied by the ratio at a time t � T, where T is the cell cycleperiod determined by Fourier analysis. If the data for a gene repeatswith a period T, the autocorrelation will be high.

A � � ratio�t�*ratio��t � T�%n� (5)

Autocorrelation scores were calculated for Thy-Thy 3 and the Thy-Noc experiments because they represent multiple cell cycles andbecause points were taken at equally spaced intervals throughouteach time course. In experiment 3 autocorrelations were calculatedfor periods (T, eq. 5) of 15, 16, and 17 h. In experiment 4 autocor-

Table 1. Periodicity scores of selected genesSample periodicity scores for representative cell cycle regulated genes. The cutoff for a cell cycle regulated gene was established by

balancing the false positive and false negative rates. We used a cutoff that gave an estimated 0.15–0.75% false positive rate and 6.5% falsenegative rate, resulting in 1333 clones. The maximum fold change is the maximum amplitude observed in any of the five synchronizations.The average fold change is the average amplitude across all five synchronizations.

Rank GenePeriodicity

scoreMax foldchange

Avg. foldchange

1 STK15 58.8 16 102 PLK 56.0 11 8.67 CENPE 42.6 11 6.817 CKS2 35.1 11 5.525 CCNA2 31.2 5.9 4.543 CDC6 25.3 3.8 3.050 CCNB1 24.3 7.3 4.154 E2F1 23.0 4.9 3.995 PCNA 16.9 3.4 2.7105 CIT 16.2 7.0 4.0135 GOLGIN-67 13.6 4.8 3.1182 ORC 1 11.2 3.7 2.2355 PRIM1 7.26 2.6 2.1499 VCL 5.80 3.4 2.7607 BRCA1 5.17 2.5 2.1618 CCNE2 7.18 3.1 2.51127 PSEN1 3.65 2.1 1.81333 CDC42 3.29 2.3 2.01334 cutoff 3.286 — —1358 ZNF162 3.25 1.9 1.81780 YY1 2.75 3 2.3

Gene Expression in the Human Cell Cycle

Vol. 13, June 2002 1979

relations were calculated for periods of 16, 18, and 20 h. The scorefor each gene in a given time course was taken to be the maximumof the three autocorrelations. The final autocorrelation score as-signed to each gene was the sum of the scores calculated for each ofthe two time courses.

Autocorrelations were used as a filter to remove genes thatshowed transient expression despite receiving a high periodicityscore. 199 Genes with a negative autocorrelation (a negative auto-correlation indicates the measured ratios do not repeat every cellcycle) were eliminated from the initial set of putative cell cycle-regulated genes. Autocorrelations were also calculated for datarandomized in rows, whereupon few genes received negative au-tocorrelations in the randomized data, indicating that the negativescores are unlikely to occur by chance. The distribution of autocor-relation scores is shown in Supplemental Figure 16.

Our final list contains 1134 clones that correspond to 874 UNI-GENE clusters (UNIGENE build 143, released November 9, 2001, 21clones not found in UNIGENE, 66 map to more than one UNIGENEcluster). The data for all 1134 clones as well as the primary data areavailable at http://genome-www.stanford.edu/Human-CellCycle/HeLa/.

RESULTS

The sensitivity with which one can detect periodic activitiesin synchronized cell cultures depends almost entirely on thedegree to which cells can be synchronized. In particular,cells that fail to begin cycling promptly upon release willcontribute noise in the analysis. We chose the well-studiedepithelial cell line, HeLa S3, derived from a cervical carci-noma (Puck et al., 1956), specifically because a high degree of

synchrony can be achieved with diverse methods (Knehr etal., 1995; Whitfield et al., 2000). HeLa cells can be synchro-nized so that 95% of the cells reenter the cell cycle, whereasthe comparable figure for primary cells (such as foreskinfibroblasts or mammary epithelial cells) is usually reportedto be between 60 and 80% (Tobey et al., 1988; Fonagy et al.,1993; Stampfer et al., 1993). Three different methods wereused to obtain synchronous populations; in each case syn-chrony was monitored either by flow cytometry or BrdUincorporation (Figure 1). A double thymidine block was firstused to arrest cells at the G1/S boundary (Adolph andPhelps, 1982), providing the best synchrony in G2 and Mphases of the cell cycle (Figure 1A). This protocol provideda somewhat less robust synchrony of the G1-to-S transition,because cells tend to differ in the rate at which they passthrough G1. To improve our resolution in the G1-to-S pe-riod, we used another method, in which HeLa cells werefirst arrested by thymidine, released, and subsequently ar-rested in mitosis with the antimicrotubule drug nocodazole(Figure 1B) (Zieve et al., 1980). In this case, better synchronywas obtained in G1 and S phases, because the synchronousrelease was at M. The final synchronization method in-volved no drugs; instead, HeLa cells at metaphase werecollected by a physical method, mitotic shake-off (Schneider-man et al., 1972), by using an automated cell shaker (Eliassenet al., 1998). More than 95% of the cells collected by mitoticshake-off reentered the cell cycle and progressed into Sphase as monitored by incorporation of BrdU (Figure 1C). In

Figure 1. Synchrony of HeLa cells. (A) Cells were arrested at the beginning of S phase by using a double thymidine block, and cellsynchrony was monitored by flow cytometry of propidium iodide-stained cells. Flow cytometry data were collected for each of the threeindependent double thymidine blocks performed in this study; data are shown only for the second double thymidine arrest (Thy-Thy2),although equivalent synchrony was obtained in each of the three experiments. The number of cells (arbitrary units) is plotted against DNAcontent for time points at 4-h intervals for 44 h; an arrow indicates the time of mitosis, as estimated from the flow cytometry data. Uponrelease from the thymidine block, �95% of the cells progressed into S phase (0–4 h), entered G2 phase (5–6 h), underwent a synchronousmitosis at 7–8 h, and reentered S phase after completing one full cell cycle at 14–16 h. Typically two to three additional synchronous cellcycles were obtained. (B) Cells were arrested in mitosis by blocking first in thymidine followed by release and then blocking in nocodazole.After release from the nocodazole block, most of the cells (�75%) divided synchronously within 2 h of release from the arrest, entered S phaseby 10–12 h after release, and completed the next synchronous mitosis by 18–20 h, ultimately completing two full cell cycles. (C) Cells wereselected by mitotic shake-off, and the percentage of cells in S phase was measured by BrdU incorporation. The percentage of cells in S phaseis plotted against the time after plating; �95% of the cells entered S phase by 12 h.



all cases, with the exception of the mitotic shake-off, two tothree synchronous cell cycles were obtained.

Identification of Cell Cycle-regulated TranscriptsTo identify cell cycle-regulated transcripts, RNA was iso-lated from HeLa cells at intervals (typically 1–2 h) afterrelease from a synchronous arrest (Figure 2A). Cy5- or Cy3-labeled cDNA was synthesized using standard protocolsand hybridized to cDNA microarrays containing either22,692 features representing �16,332 different human genes(Thy-Thy 1 and Thy-Thy 2) or 43,198 features representing�29,621 different genes (Thy-Thy 3, Thy-Noc, and MitoticShake-off) as estimated by UNIGENE clusters.

The data from each of the synchronization experimentsrequired separate analysis before combination into a singledataset as described in MATERIALS AND METHODS. Toidentify those transcripts that were cell cycle regulated, aperiodicity score was obtained for each clone using a Fouriertransform and correlation to known cell cycle genes as de-scribed for yeast (Spellman et al., 1998). Representative genesand their periodicity scores are shown in Table 1; the rangewas 0–58.8.

The minimum score for a gene we designated as “cellcycle-regulated” was determined by estimating the falsepositive rate by using randomized data as described inMATERIALS AND METHODS. We chose a threshold scoreof 3.29 that gave an estimated 10 false positives (0.75%;periodicity scores of 5.18–3.33) when the data were random-ized only within rows (i.e., within each gene), and two falsepositives (0.15%; periodicity scores of 3.72–3.30) when rowsand columns were randomized. Using this relatively conser-

vative threshold, we identified 1333 clones as cell cycleregulated.

A fraction of the genes with relatively high Fourier scoresvaried in a quasi-sinusoidal manner during part of the timeseries but clearly were not periodically expressed. This quasi-sinusoidal variation resulted from variation in gene expres-sion at the beginning of each time course, probably as aresult of the synchronization procedure. Possible sources ofthis variation are the serum response, resulting from theaddition of fresh growth media upon release of the cellsfrom the arrest, or simply a stress response resulting fromthe cell cycle block. To filter out genes that did not oscillateacross multiple cell cycles, each clone was assigned an au-tocorrelation score that describes whether the expressionratio at a given time is a good predictor for the value of theexpression ratio one cell cycle later. Thus, genes with aconsistent pattern of periodic expression every cell cyclehave positive autocorrelation scores, whereas genes withlow or negative scores do not have consistent periodic ex-pression. Using this method, 199 of the initial 1333 clonesreceived negative autocorrelation scores and were removedfrom the list, leaving 1134 clones representing 874 differentgenes.

To assess the quality of the criteria for a periodicallyexpressed gene, the false negative rate was estimated byusing a list of known cell cycle-regulated genes compiledfrom the literature. The known genes were limited to thoseregulated at the mRNA level during a continuous humancell cycle, as determined by traditional methods (e.g., North-ern blot, S1 nuclease assay, or RNase protection; Table 2).Many “known” cell cycle genes are not included in this list

Figure 2. Gene expression of well-characterized cell cycle genes. The expression profiles are shown for 20 well-characterized genes whoseexpression peaks in each phase of the cell cycle. Data are shown for each of the five time courses (114 total arrays). The purple bars aboveeach column represent S phase and the arrows indicate mitoses as estimated by flow cytometry or BrdU labeling. In each time course, RNAwas collected for points (typically every 1–2 h) for 30 h (Thy-Thy1), 44 h (Thy-Thy2), 46 h (Thy-Thy3), 36 h (Thy-Noc), or 14 h (Shake) afterthe synchronous arrest, at the intervals indicated. The four genes representing each cell cycle phase were averaged to create five different“ideal expression profiles” (G1/S, S, G2, G2/M, and M/G1).


Vol. 13, June 2002 1981

because their regulation has been demonstrated only duringthe resting-to-growing transition, not in cells synchronizedindependent of serum, or because data for the regulation ofthe mRNA could not be found, despite a sometimes exten-sive literature on protein levels, localization, and functionduring the cell cycle (e.g., CENPE; Wood et al., 1997; Abrieuet al., 2000). Of the 49 known cell cycle-regulated genes inour list, three were not measured in our analysis. Of theremaining 46, 52.2% (24/46) are found in the top 120 scoringclones, and 93.5% (43/46) are included in the top 850 clones,yielding a false negative rate of 6.5%. Although most of theknown genes are represented in the top 850 clones, webelieve there is still significant information to be obtainedbetween our 850 and 1134 clones, which include duplicateclones of many of the known genes (e.g., the histones andRAD51) and many genes expressed during mitosis.

Classification of Transcripts by Cell Cycle PhaseFigure 2 presents data for well-studied genes that showedpeak expression at specific points in the cell cycle. The timesof S phase and mitosis, as estimated from the flow cytometryand BrdU data (Figure 1), are indicated in Figure 2. Maxi-mum expression of the mitotic cyclins (CCNE1, CCNA2,and CCNB1) was observed in the known temporal order andat the expected times (Pines and Hunter, 1989; Pines andHunter, 1990; Lew et al., 1991). Cyclin E1 expression at theG1/S boundary was accompanied, as expected, by E2F1 andthe DNA replication factors CDC6 and PCNA (Morris andMathews, 1989; Ohtani et al., 1995; Yan et al., 1998). ManyDNA metabolism genes were expressed at the beginning ofS phase; these are represented here by RFC4, DHFR, RRM2,and RAD51. In our data, two distinct groups of mitotic genescould be distinguished. Transcripts of some genes, such asCCNA2, TOP2A, Cyclin F, and CDC2, peaked in G2,whereas others, such as CCNB1, BUB1, STK15, and PLK1,peaked in mitosis. The differences in expression of the G2and M phase genes are most evident in the third doublethymidine arrest experiment in which samples were takenevery hour for 46 h and at the earliest times after the thy-midine-nocodazole arrest (Figure 2). We have followed ourdata, rather than precedent, in recognizing that many, if notall, the genes expressed during the physical act of mitosiscontinue to be expressed into the G1 period and have thuslabeled these genes M/G1. These genes are represented inFigure 2 by RAD21, PTTG1, VEGFC, and CDKN3.

All 1134 clones identified as periodically expressed weresorted by the point in the cell cycle when they showed peakexpression as calculated from the sine and cosine compo-nents for the Fourier transform (Figure 3A). The periodicnature of the expression patterns is evident from the alter-nating red (strong expression relative to the asynchronousreference) and green (weak expression relative to the asyn-chronous reference). It is notable that these patterns persistacross multiple cell cycles in both Figures 2 and 3.

To minimize unnecessary reassignment of genes relativeto the published literature, each gene was assigned a cellcycle phase (G1/S, S, G2, G2/M, or M/G1) by their peakcorrelation to an idealized expression profile generated fromthe well-studied genes selected for Figure 2. These assign-ments are necessarily somewhat arbitrary because a geneassigned to G1/S can be immediately adjacent to a geneassigned to S phase. By assigning phase in this way, 211

(18.6%) of the clones were maximally expressed during theG1/S transition, and 221 (19.5%) were maximally expressedin S phase. Genes in each of these phases have a known rolein replication initiation and DNA metabolism. More than50% of the clones showed peak expression in G2 and Mphases of the cell cycle; 239 clones (21.1%) peaked in G2, 273clones (24.1%) at G2/M, and 190 clones (16.8%) at M/G1.Genes with maximal expression levels in G2 and M phaseshave roles in chromosome metabolism, surveillance of mi-totic processes, and cell adhesion. Interestingly, relativelyfew genes show peak expression between M/G1 and G1/S,as judged from the distribution of arctangent values calcu-lated from the Fourier transform (Supplemental Figure 15).

In most cases the phase assignments we made for theknown genes are in agreement with the published literature(Table 2). Several genes that were previously characterizedto peak in a specific cell cycle phase were assigned to anadjacent phase. Examples are CDC45L and CDKN3, bothreported to peak at G1/S, whereas in our data they fell intothe S phase and M/G1 groups, respectively (Table 2). Thedifference in assignments may represent differences in ex-perimental systems, methods of measurement, or resolutionin the different experiments.

The replication-dependent human histone mRNAs are notpolyadenylated (Marzluff, 1992) and thus behave sporadi-cally in all experiments except the mitotic shake-off experi-ment where total RNA was labeled. Because of this incon-sistent behavior, they were incorrectly assigned a phase ofG1/S by our method. Because the data obtained from mi-totic selection and extensive literature on the cell cycle reg-ulation of the histone genes indicated they are expressed inS phase, the phase has been assigned manually, rather thanby correlation to the ideal vector (indicated by an asterisk [*]in Table 2 and the supplemental data).

Overview of Periodically Expressed Gene ClustersHierarchical clustering of transcripts with similar expressionpatterns often groups genes according to the processes inwhich they participate; these processes can often be inferredfrom the annotations of the known genes in each group(Eisen et al., 1998). To begin to identify the processes that arerepresented in our clusters, we annotated the known genesin our clusters with gene ontology (GO) terms (Ashburner etal., 2000) for biological process from LocusLink (Pruitt andMaglott, 2001). GO, initially developed using the modelorganisms S. cerevisiae, Drosophila melanogaster, and themouse Mus musculus, provides a controlled vocabulary todescribe the functions of known genes. GO terms are di-vided into three different categories: 1) molecular function,2) biological process, and 3) cellular component. Examina-tion of the lists of clustered genes and their annotationsshows that within our set of 1134 clones there are clusters ofgenes that have similar or identical annotations (Figure 3B).In what follows, GO terms are italicized and terms we haveapplied provisionally herein because of the incomplete ap-plication of GO to human genes are marked with an asterisk(*). The DNA replication clusters include genes associatedwith the following process annotations: DNA replication, ex-pressed in G1/S (early DNA replication; e.g., the compo-nents of the prereplication complex and ORC1); DNA metab-olism (including its daughter processes DNA repair and DNArecombination) and nucleotide metabolism, expressed in S (late



Table 2. Known cell cycle-regulated genesA list of genes was compiled from the literature that had been shown to be cell cycle regulated by traditional methods in cells synchronized

independent of serum stimulation. In each case we have listed the primary reference describing the cell cycle regulation of the gene and thepredicted cell cycle phase. The cell cycle phase assigned in our study is shown on the right. *, histone genes were inappropriately assignedto G1/S because they are not polyadenylated and hence behaved sporadically in all experiments except the mitotic selection experimentswhere total RNA was labeled. Due to extensive literature on the cell cycle regulation of the histone genes and their S phase expression in themitotic shake-off experiment, the phase was reassigned manually to S phase.

Gene symbol AccessionPublished

phase ReferenceAssigned

phase

E2F5 NM_001951 G1 (Sardet et al., 1995; Grosheva et al., 2001) G2/MCCNE1 M74093 G1/S (Lew et al., 1991) G1/SCCNE2 NM_004702 G1/S (Zariwala et al., 1998) G1/SCDC25A NM_001789 G1/S (Jinno et al., 1994) G1/SCDC45L NM_003504 G1/S (Saha et al., 1998) S phaseCDC6 NM_001254 G1/S (Leone et al., 1998; Yan et al., 1998) G1/SCDKN1A, p21 NM_000389 G1/S (Hiyama et al., 1998) NDCDKN3 NM_005192 G1/S (Gyuris et al., 1993) M/G1E2F1 NM_005225 G1/S (Leone et al., 1998) G1/SMCM2 NM_004526 G1/S (Leone et al., 1998) G1/SMCM6 NM_005915 G1/S (Leone et al., 1998) G1/SNPAT NM_002519 G1/S (Ma et al., 2000; Zhao et al., 2000) G1/SPCNA NM_002592 G1/S (Morris and Mathews, 1989) G1/SSLBP NM_006527 G1/S (Whitfield et al., 2000) G1/SBRCA1 NM_007295 S phase (Rajan et al., 1996) S phaseBRCA2 NM_000059 S phase (Rajan et al., 1996; Vaughn et al., 1996) NDCCNG2 NM_004354 S phase (Horne et al., 1996) NDCDKN2C NM_001262 S phase (Okuda et al., 1995) G2DHFR NM_000791 S phase (Slansky and Farnham, 1996) S phaseHistone H1 multiple genes S phase (Heintz et al., 1983; Plumb et al., 1983; Harris et al., 1991) S phase*Histone H2A multiple genes S phase (Heintz et al., 1983; Plumb et al., 1983; Harris et al., 1991) S phase*Histone H2B multiple genes S phase (Heintz et al., 1983; Plumb et al., 1983; Harris et al., 1991) S phase*Histone H3 multiple genes S phase (Heintz et al., 1983; Plumb et al., 1983; Harris et al., 1991) NDHistone H4 multiple genes S phase (Heintz et al., 1983; Plumb et al., 1983; Harris et al., 1991) S phase*MSH2 NM_000251 S phase (Marra et al., 1996) G1/SNASP NM_002482 S phase (Richardson et al., 2000) G1/SRRM1 NM_001033 S phase (Bjorklund et al., 1990) S phaseRRM2 NM_001034 S phase (Bjorklund et al., 1990) S phaseTYMS NM_001071 S phase (Kaczmarek, 1986) S phaseCCNA2 NM_001237 G2 (Pines and Hunter, 1990) G2CCNF NM_001761 G2 (Bai et al., 1994) G2CENPF NM_016343 G2 (Liao et al., 1995) G2/MTOP2A NM_001067 G2 (Goswarni et al., 1996) G2BIRC5 NM_001168 G2/M (Li and Altieri, 1999) G2/MBUB1 NM_004336 G2/M (Ouyang et al., 1998) G2/MBUB1B NM_001211 G2/M (Davenport et al., 1999) G2/MCCNB1 NM_031966 G2/M (Pines and Hunter, 1989) G2/MCCNB2 NM_004701 G2/M (Jackman et al., 1995) G2/MCDC2 NM_001786 G2/M (Dalton, 1992; Zwicker et al., 1995) G2CDC20 NM_001255 G2/M (Weinstein et al., 1994; Weinstein, 1997) G2/MCDC25B NM_021874 G2/M (Jinno et al., 1994) G2/MCDC25C NM_001790 G2/M (Lucibello et al., 1995; Zwicker et al., 1995) G2CDKN2D, p19 NM_001800 G2/M (Hirai et al., 1995) G2/MCENPA NM_001809 G2/M (Shelby et al., 1997) G2/MCKS1 BC001425 G2/M (Richardson et al., 1990) G2CKS2 NM_001827 G2/M (Richardson et al., 1990) G2/MPLK NM_005030 G2/M (Holtrich et al., 1994; Lee et al., 1995) G2/MSTK15 NM_003600 G2/M (Shindo et al., 1998) G2/MSTK6 NM_003158 G2/M (Kimura et al., 1997) NMRACGAP1 NM_013277 G2/M (Hirose et al., 2001) NMRAB6KIFL NM_005733 G2/M (Fontijn et al., 2001) NM

ND, not detected as cell cycle regulated; NM, not measured in our microassays.


Vol. 13, June 2002 1983

Figure 3. Gene expression of the human cellcycle. (A) The 1134 periodically expressedclones are ordered by the point in the cellcycle where they show peak expression, ascalculated from the sine and cosine coeffi-cients of the Fourier transform. Each genewas assigned a cell cycle phase based on itscorrelation to an idealized expression profiledefined in Figure 2. Color bars on the rightrepresent the phase in which the majority ofthe genes are expressed (G1/S, green; Sphase, blue; G2, red; G2/M, orange; andM/G1, black); this color scheme is usedthroughout the remainder of the figures. (B)The 1134 clones that have similar expressionprofiles are grouped together by hierarchicalclustering. The biological theme of each clus-ter is shown on the right, and the color of thebar indicates the cell cycle phase representedby the genes. The cell adhesion and chromo-some metabolism clusters have been labeleda, b, and c as each cluster contains similar GOannotations. Both figures, with all genenames and accession numbers, are availableas Supplemental Figure 9 at the Mol. Biol.Cell Web site and at a Web site maintained bythe authors (http://genome-www.stanford.edu/Human-CellCycle/HeLa/).



DNA replication; e.g., RAD51, DNA polymerases, and nu-cleotide metabolism enzymes); and chromatin assembly/disas-sembly (e.g., the histones), also expressed in S. The mitoticclusters include mitotic chromosome segregation (the �- and

�-tubulins) expressed in G2; spindle assembly, expressed dur-ing G2 (e.g., kinesins and centrosome duplications genes);mitotic checkpoint and nuclear division, expressed at G2/M(e.g., the checkpoint genes BUB1 and CENPE); sister chroma-

Figure 3 (legend on facing page).


Vol. 13, June 2002 1985

Figure 4.



tid cohesion (e.g., PTTG1 and genes of the cohesin complexSTAG1, RAD21); and cell adhesion (e.g., CTNND1 and vin-culin) and components of the RAS signal transduction path-way expressed during the physical act of mitosis, M/G1.The first point to be made about these clusters (shown inmore detail in Figures 6–8 and the supplemental data,which contain the complete cluster diagram with genenames) is that each represents a reasonably complete andspecific list of the genes necessary for the major, essentialprocesses that must occur every cell division cycle. Thesecond point to be made about these is that, in addition togenes of known function, there are numerous genes of here-tofore uncharacterized function that are now implicated inthese processes.

Tumor Proliferation ClusterRecently, several studies of global gene expression patternsin human tumors and tumor cell lines have been published(Perou et al., 1999, 2000; Alizadeh et al., 2000; Ross et al.,2000). Each of these studies identified a prominent cluster ofgenes whose expression was correlated with the rate ofproliferation of the tumors or cell lines under study. As weshow now, half or more of the genes that comprise this“proliferation cluster” are the same genes that we havefound to be periodically expressed in HeLa cells.

To compare explicitly the genes of the proliferation clusterwith the 874 periodically expressed genes described here,we took the list of genes in each of the proliferation clustersidentified in the studies of breast tumors (Perou et al., 2000)and lymphoma (Alizadeh et al., 2000), and extracted thepatterns of gene expression for each from the dataset de-scribed here. These patterns, lined up by peak expressionduring the cell cycle as in Figure 3, are presented for 112genes measured here and in breast cancer (Figure 4A) and

for 96 genes measured here and in lymphoma (Figure 4B).As the figures show, 62% (69/112) of the genes in the breasttumor proliferation cluster and 45% (43/96) of the lym-phoma proliferation cluster and are among the 874 period-ically expressed genes we detected in synchronized HeLacell cultures.

The simplest interpretation of this result is that the geneswhose transcript levels correlate with the proliferation rateof the tumors are expressed only in cycling cells, presum-ably as a consequence of regulation that allows transcriptiononly during the appropriate stage of the cell cycle and not atany other time, including quiescence. Even if this idea iscorrect, it need not apply to all genes that are periodicallyexpressed during the cell cycle. It seems quite possible thatsome genes that are periodically expressed in cycling cellsmight not appear in the proliferation clusters of tumorsbecause they might be expressed under some circumstancesin noncycling cells or strongly regulated not only by the cellcycle but also by related factors, and thus show little asso-ciation with tumor proliferation. One might even anticipatethat some periodically expressed genes might be regulatedin ways (e.g., strong expression in differentiated nonrepli-cating cells) that result in reduced transcription during ac-tive cell division cycles.

To explore further the relationship between periodic ex-pression during the cell cycle and expression in proliferativetumors, we have extracted the data for every clone that wasboth identified as periodically expressed in this study andalso measured in the breast tumor study, regardless of itspattern there (Perou et al., 2000). This yielded 386 clones, ofwhich �25% appeared in the proliferation cluster. Hierar-chical clustering of this set of 386 using the combined datafrom both studies revealed, as anticipated, not only period-ically expressed genes that are highly correlated with tumorproliferation (a more comprehensive proliferation cluster, asdescribed above; Figure 5B, ii and iii) but also periodicallyexpressed genes that show heterogeneous expression in tu-mors (i.e., no obvious association with tumor proliferation;Figure 5B, v), and periodically expressed genes whose ex-pression is apparently uncorrelated with the proliferativestate of the tumors (Figure 5B, i and iv).

Correlation between periodic expression in the cell cycleand strong expression in highly proliferative tumors can beaccounted for simply because there is an increase in thenumber of cycling cells (as described above), but other morespecific explanations can also be proposed. Many genes thatwe have found to be periodically expressed have alreadybeen implicated directly in tumorigenesis and their expres-sion has been reported to coincide with cellular transforma-tion and oncogenic potential (reviewed in Bishop, 1991;Hunter, 1997). Genes of this kind are common in clusterswhose expression is highly correlated with tumor prolifer-ation in our study, including some expressed primarily inthe G2 and M phases (Figure 5B, ii) and others whoseexpression peaks in G1 and S phases (Figure 5B, iii).

Among the periodically expressed genes whose expres-sion is heterogeneous or uncorrelated with tumor prolifera-tion are genes for a variety of cell cycle processes, includinggenes for cell-cell adhesion (calponin-2, smoothelin, andvinculin) (Figure 5B, i). Expression of some of these genes,such as vinculin, in transformed cells has been reported todecrease tumorigenecity (Rodriguez Fernandez et al., 1992).

Figure 4 (facing page). Genes associated with the proliferativestate of tumors are cell cycle regulated. (A) Cell cycle regulation ofthe breast tumor proliferation cluster. The expression patterns of the112 clones in the breast tumor proliferation cluster (Perou et al.,2000) were examined in the cell cycle dataset, along with theirexpression in breast tumors and cell lines. The clones are ordered bytheir peak expression during the cell cycle (as in Figure 3A) suchthat each row corresponds to a single clone. The order of the breasttumor samples based on hierarchical clustering by using 1753 genesis as described previously (Perou et al., 2000). The mitotic grade ofthe tumors is indicated (a higher mitotic grade is indicative of amore proliferative tumor). Of the 112 genes in the proliferationcluster, 69 were identified as cell cycle regulated in this study. (B)Cell cycle regulation of lymphoma proliferation cluster. Because thelymphoma study was performed on the Lymphochip Microarray,which is comprised of a different clone set than the Stanford HumanMicroarray used for this study, the genes on the two arrays werelinked via their LocusLink ID (Pruitt and Maglott, 2001). We se-lected 154 clones from the previously published lymphoma prolif-eration cluster (Alizadeh et al., 2000), and 19 clones without Lo-cusLink IDs were discarded. Of the remaining 135, 128 clones withmatching LocusLink IDs were measured in our cell cycle study.Duplicate clones were removed leaving 96 genes. Of these, 43 wereidentified as cell cycle regulated in this study. As above, the genes(i.e., rows) are ordered by peak expression during the cell cycle. Theorder of the tumors, based on hierarchical clustering by using alarger data set, is as published previously (Alizadeh et al., 2000).This figure with all gene names is available as Supplemental Figure10.


Vol. 13, June 2002 1987

Another group of cell cycle-regulated genes that are notcorrelated with tumor proliferation includes genes necessaryfor withdrawal from the cell cycle and apoptosis (the cyclin-dependent kinase [cdk] inhibitor CDKN2C (p18) andCaspase 3; Figure 5B, iv). Finally, a group of G1 and S phasegenes shows heterogeneous expression in tumors (Figure5B, v), including DHFR, E2F1, CDC6, CCNE1, CHAF1A,TOP3A, ORC1, and BRCA1; these data suggest that theregulation of each of these genes is more complex thansimple restriction of transcription to a particular phase of thecell division cycle.

Biological Roles of Periodically ExpressedGene ClustersThe close similarity between the set of genes whose expres-sion is associated with tumor proliferation and those that areperiodically expressed invites a closer examination of thebiological roles of the gene clusters detected in Figure 3B, inlight of the specific features of both the cell cycle and tumorbiology. In what follows, we survey roles of the genes ofknown function in each of the prominent clusters. It shouldbe recalled, however, that each of these clusters contains, in

Figure 5. Expression of cell cycle-regulated genes in breast tumors. The tumor samples have been ordered by hierarchical clustering using1753 genes as described previously (Perou et al., 2000). (A) Expression profiles are shown of 386 of the 1134 cell cycle-regulated clones thatwere also present in the breast cancer study (Perou et al., 2000). The clones are ordered by their similarity of gene expression profiles acrossboth data sets by hierarchical clustering. (B) Detailed views of the clusters identified in Figure 5A. A subset of genes is expressed at lowerlevels in the more proliferative tumors (NODE379X, cluster i; NODE303X, cluster iv) and includes genes that peak at M/G1, S, and G2 phases.Genes for core cell cycle processes that peak at G2 and M phases (NODE142X, ii) or at G1 and S phases (NODE133X, iii) are highly expressedin the more proliferative tumors; many of these genes are the same genes found in the breast tumor proliferation cluster. A subset of cell cycleregulated genes is expressed in a heterogeneous manner (NODE345X, v). An enhanced figure with all gene names is available asSupplemental Figure 11.



addition to the genes of known function, genes whose bio-logical role remains to be determined (expressed sequencetags; ESTs). The numbers of such uncharacterized genes arenoted in the legends to Figures 6–8 and, of course, thedetailed position of each of them can be found in the sup-plementary information and on the Web site.

G1 and S Clusters

The “early DNA replication” cluster contains genes ex-pressed in late G1 phase with their expression continuinginto S phase. The GO process annotations that apply to mostof these genes are DNA replication (including well-character-

Figure 5 (legend on facing page).


Vol. 13, June 2002 1989

Figure 6.



ized genes encoding RFC4, DNA polymerase delta 3,DDX11, and geminin), DNA packaging (including CHAF1A,CHAF1B, and PCAF), and DNA repair (including MSH2,FEN1, and PCNA). Many of the genes in the DNA replica-tion category are components of prereplicative complex andinclude CDC6, ORC1L, MCM2, MCM4, MCM5, and MCM6(Lei and Tye, 2001); this group of genes is consistentlyexpressed at relatively high levels in diverse types of tumorsand has been proposed as diagnostic markers by Laskey andcoworkers (Williams et al., 1998). The cluster also includesgenes that fall into the GO category cell cycle control and arenecessary for entry into S phase (in this group are Cyclin E1,E2F1, CDC25A, and Cyclin E2).

The “late DNA replication” cluster (Figure 6B) containsgenes necessary for the continued synthesis of DNA, includ-ing nucleotide metabolism (including TYMS, DHFR, RRM1,and RRM2) and additional DNA replication genes (DNApolymerase alpha and theta; Primase 1 and 2A). Alsopresent are genes for DNA repair and DNA recombination(RAD54 and RAD51). Genes involved in DNA repair werealso expressed in S phase in a study of the cell cycle inprimary human fibroblasts, and it was reported that thesegenes were also induced when cells were treated with UVradiation (Cho et al., 2001).

Histone synthesis is required during DNA replication topackage newly synthesized chromatin and is tightly regu-lated during the cell cycle (Marzluff and Pandey, 1988;Schumperli, 1988). The replication-dependent histonegenes are coordinately regulated and the expression of theirmRNAs is tightly restricted to S phase of the cell cycle byboth transcriptional and posttranscriptional mechanisms(Harris et al., 1991; Eliassen et al., 1998). The “histone” cluster(Figure 6C), expressed during S phase, contains the corehistones H2A, H2B, and H4 and the linker histone H1, whichwe have assigned to the GO category of chromatin assembly/disassembly. These mRNAs are not polyadenylated andhence are measured poorly and inconsistently in all experi-ments except mitotic selection where total RNA was labeled.It should be noted that the several H2A genes will cross-hybridize, as will the H2B, H4, and H1 genes, but significantcross-hybridization between the different genes (e.g., H2Aand H2B) is unlikely. Histone H3 mRNA, although cellcycle-regulated, was not found in our study, possibly be-

cause of preferential labeling of the polyadenylated, non-cellcycle-regulated variant H3.3 (Wells and Kedes, 1985).

Two regulators of mammalian histone mRNA synthesisare present in the G1/S cluster (Figure 6A), the stem-loopbinding protein (SLBP) necessary for histone pre-mRNAprocessing (Wang et al., 1996a; Dominski et al., 1999), andnuclear protein mapped to the AT locus (NPAT), which wasrecently identified as an activator of histone gene transcrip-tion and is hypothesized to be a chromatin remodelingfactor (Ma et al., 2000; Zhao et al., 2000). The SLBP waspreviously shown to be cell cycle regulated by both transla-tional and posttranslational mechanisms; the magnitude ofthe changes observed herein in the mRNA level (approxi-mately twofold) is consistent with Whitfield et al. (2000).

Many of the genes expressed at G1/S and S phase areknown E2F targets. A study of the cell cycle and the E2Ftranscription factors in mouse embryo fibroblast, using mi-croarrays, identified both G1/S and S phase genes, as well asgenes expressed at G2 and M phases, as targets of the E2Ftranscription factor (Ishida et al., 2001). Ishida et al. (2001)identified the histone SLBP as induced by E2F, suggestingthat histone synthesis may be linked to the Rb/E2F path-way. Furthermore, NPAT is a cyclin E/cdk2 substrate andits overexpression promotes S phase entry (Zhao et al., 1998).Although NPAT itself has not been implicated in tumori-genesis, the role of chromatin remodeling in cancer is be-coming increasingly clear (Archer and Hodin, 1999).

G2 and M ClustersThe mitotic genes form three distinct clusters, one of whichincludes genes that encode �- and �-tubulins; the other twocontain well-characterized genes that encode proteins in-volved in assembly and surveillance of the mitotic spindle(Figure 3B). The remaining G2 and M phase genes that falloutside of these core clusters also have functions concernedwith the GO processes mitosis, and its daughter processes(e.g., mitotic chromosome segregation, spindle assembly, mitoticcheckpoint, etc).

The tubulin cluster includes �- and �-tubulins (TUBA1,TUBA2, TUBA3, TUBB, and TUBB2) as well as BUB3, whichis a mitotic checkpoint regulator (Figure 7A). (Note thatthere may be cross-hybridization among the several �-tubu-lin genes or between the several �-tubulin genes, but cross-hybridization between �- and �-tubulin genes is unlikely.)Coordinate regulation of tubulin synthesis at the level ofmRNA abundance is evident herein, even though a largecomponent of the regulation of these genes is posttranscrip-tional (Cleveland, 1989). Yeast BUB3 was isolated as a mul-ticopy suppressor of a conditional mutation in �-tubulin(Guenette et al., 1995), the coexpression of hBUB3 with thecoordinately regulated tubulins is consistent with its role ofmonitoring microtubule assembly as cells enter mitosis.

The second G2 phase cluster contains genes predomi-nately for organization of the mitotic spindle, includingkinesins and the TTK kinase (human homolog of the S.cerevisiae MPS1 spindle checkpoint gene; Figure 7B; Abrieuet al., 2001). Five different kinesins are present: KNSL1,KNSL2, KNSL4, KNSL5, and KNSL6. KNSL6 is expressed atthe boundary between G2 and M phase genes and in ourstudy has been assigned a phase of G2/M. In addition toCyclin A2, CDC2, CKS1, Cyclin F, and numerous otherregulatory factors are found in this cluster. Also specifically

Figure 6 (facing page). G1 and S phase clusters. Shown are de-tailed views of the cell cycle clusters indicated in Figure 3B. Only aportion of each cluster is shown; the full cluster with all gene names,phase assignments, and annotations can be viewed in SupplementalFigure 12. The names of selected genes and their correspondingrows (indicated by an arrow) are shown to the right. Some geneshad multiple clones on the arrays that clustered together (e.g.,CDC25A and Golgin-67) and are indicated by a bracket. The GObiological process annotation is shown for each gene. Due to theincomplete application of GO to the human genome, we haveassigned the biological process to some genes provisionally; anasterisk (*) precedes these terms. (A) Early DNA replication cluster(NODE1077X) contains 202 clones (94 ESTs and hypothetical pro-teins); 88% have been assigned a phase of G1/S. (B) Late DNAreplication cluster (NODE793X) contains 139 clones (66 ESTs andhypothetical proteins); 80% show peak expression in S phase. (C)Histone gene cluster (NODE 516X) contains 13 clones expressedduring S phase; 12 of the clones are histone genes.


Vol. 13, June 2002 1991

expressed in G2 is ESP1 or human Separin, which is releasedfrom PTTG1 (human homolog of the yeast securin PDS1)during anaphase and promotes chromatid separation (Zouet al., 1999). Finally, it may be worth noting that Importinalpha, also in this cluster, has recently been shown to inhibitspindle assembly by binding a protein that promotes spindleformation, Repp86 (Xenopus TPX2) (Gruss et al., 2001; Na-chury et al., 2001; Wiese et al., 2001). This inhibition is re-lieved by the action of the small GTPase RAN when boundto GTP (present in the M/G1 cluster; Figure 8), which isproposed to surround M phase chromosomes (Carazo-Salaset al., 1999; Hetzer et al., 2000), thus allowing spindle forma-

tion only in the vicinity of chromosomes. The temporalsuccession of the transcripts for these two proteins suggestsa possible role in regulating the mitotic spindle.

The third mitotic cluster contains genes that peak at G2/Mand centers on Cyclin B2 (Figure 7C); Cyclin B1 falls justoutside this cluster. Genes that have known functions in themitotic spindle checkpoint show peak expression at G2/M,including BUB1, BUB1B, CDC20, and CENPE. Three genesin this cluster (STK15, PLK1, and NEK2) have been shown tohave roles in centrosome duplication, whose improper ex-pression is a potential cause of tumor aneuploidy (Lengaueret al., 1998). It is worth noting that PLK1 and STK15 had the

Figure 7. G2 and M phase clusters. Clusters of genes from Figure 3B involved in mitotic processes are shown in more detail. The annotationis as described in Figure 6. (A) Tubulin gene cluster (NODE481X) contains 18 clones (six ESTs) that peak at G2 of the cell cycle. The coordinateexpression of the �- and �-tubulins is notable. (B) Spindle assembly cluster (NODE180X) contains 49 clones (22 ESTs); 88% have been assigneda phase of G2, and many of the known genes are involved in establishment and maintenance of the mitotic spindle. (C) Mitotic surveillancecluster (NODE191X) contains 50 clones (13 ESTs) that primarily peak in G2/M. The full clusters with all gene names, phase assignments, andannotations can be viewed in Supplemental Figure 13.



very highest periodicity scores (Table 1), and thus are themost periodically synthesized mRNAs among the cell cycle-regulated transcripts.

M/G1 ClustersThree groups of genes whose expression peaks in mitosismerit a more detailed examination. We classified these genesas M/G1 because they are expressed during the physical actof mitosis and their transcripts persist into G1 phase. Someof these genes are known to have roles in cell adhesion,chromosome remodeling, and membrane trafficking as op-posed to roles specifically associated with the progress of thecell division cycle.

The M/G1 cluster contains genes known to function inactin cytoskeleton remodeling and cell adhesion (Figure 8). As

a synchronous population of cells proceeds through the cellcycle, visible changes occur to the cellular morphology. Themost notable of these changes is the dramatic remodeling ofcell shape that can be observed in cell culture during mitosis,when cells detach from the plate and surrounding cells,undergo cytokinesis, and reattach to the plate as they enterthe next cell cycle (Schneiderman et al., 1972). A cluster ofM/G1 genes is expressed during this transition and containsgenes whose products have been localized to the structuresthat connect the actin cytoskeleton to the membrane or havebeen implicated in the control of cell-cell contacts.

The Ras GTPase can induce perturbations in cell-cell con-tacts when overexpressed in its constitutively activated form(Kinch et al., 1995; Zhong et al., 1997; Yamamoto et al., 1999).Included in the M/G1 cluster is Kirsten Ras-2 (KRAS2),

Figure 8. M/G1 clusters. Three clusters of M/G1 genes from Figure 3B are shown in more detail. Each contains genes involved in celladhesion, chromosome metabolism and RAS signal transduction and is annotated as described in Figure 6. Cluster A, (NODE568X) contains104 clones (33 ESTs and hypothetical proteins); 37% show peak expression in G2/M and 63% show peak expression in M/G1. Cluster B,(NODE688X) is comprised of 12 clones (three ESTs), including CDK7, the cdk of the cdk-activating complex (CAK); 92% show peakexpression in M/G1. Cluster C, (NODE988X) which peaks in late M/G1, contains 25 clones (eight ESTs and hypothetical proteins); 96% peakin M/G1. The full clusters with all gene names, phase assignments and annotations can be viewed in Supplemental Figure 14.


Vol. 13, June 2002 1993

regularly mutated in many malignancies, including lungadencarcinoma, mucinous adenoma, ductal carcinoma ofthe pancreas, and colon tumors (Barbacid, 1990) (Figure 8A).Ras is believed to act through downstream targets, includingCDC42 (which appears in this cluster; Figure 8C) as well asthe Rho GTPases (which we do not detect as cell cycle-regulated). These GTPases regulate both the assembly ofactin filaments and the assembly of the adhesion complexes.Among the M/G1 genes that have been localized to cell-celladherens junctions are catenin delta 1 (CTTND1, also knownas p120), which regulates cadherin clustering at adherensjunctions through Rho GTPases (Anastasiadis and Reynolds,2001); presenilin 1, which binds E-cadherin at cell-cell junc-tions (Georgakopoulos et al., 1999) and has also been shownto localize to the nuclear membrane, interphase kineto-chores, and centrosomes, suggesting a function in chromo-some segregation (Li et al., 1997); vinculin, involved in at-taching actin to the membrane (Jockusch and Isenberg, 1981;Wilkins and Lin, 1982); calponin 2 (CNN2), which containsan actin binding site and is localized to adherens junctionswith vinculin (Masuda et al., 1996); and MLLT4 (known alsoas AF-6 or afadin), which is a component of tight junctions(Yamamoto et al., 1997). Taken together, these results sug-gest that the regulation of transcripts during the cell cycleplays a role in orchestrating the morphological changes thatoccur during mitosis. Expression of these genes during thephysical act of mitosis may be necessary to prepare the cellto reestablish contact with the surrounding milieu.

Genes involved in chromosome architecture and remod-eling are also expressed at M/G1 (Figure 8A). These includepresenilin 1, discussed above, and parathymosin, which islocalized to sites of early DNA replication (Vareli et al., 2000)and binds to the linker histone H1 (Kondili et al., 1996). Thedyskeratosis gene (DKC1) is implicated in telomere mainte-nance, binds to H/ACA class of snoRNPs, and has beenshown to associate with the telomerase RNA that acts as atemplate to add sequence repeats to the ends of chromo-somes (Mitchell et al., 1999). Two members of the SWI/SNFcomplex (SMARCB1 and SMARCD1), present in the M/G1cluster, function in chromatin remodeling and transcrip-tional regulation (Wang et al., 1996b; Muchardt and Yaniv,2001). Other M/G1 genes with roles in chromosome archi-tecture and remodeling include human securin (PTTG1),and RAD21, both genes involved in sister chromatid cohe-sion (Zou et al., 1999; Hoque and Ishikawa, 2001).

Finally, expression of CDK7 (MO15) the catalytic subunitof the CDK-activating kinase (CAK) also peaks at M/G1(Figure 8B). CDK7 is a catalytic subunit of CAK, necessaryfor the phosphorylation and full activity of CDK2 and CDC2(Fisher and Morgan, 1994). CAK activity and CDK7 proteinhave been reported to be constant during the cell cycle(Matsuoka et al., 1994; Tassan et al., 1994) although data onthe mRNA regulation have not previously been reported.The CAK cyclin subunit Cyclin H (Fisher and Morgan, 1994)does not oscillate in our dataset even though we do seeoscillation of the mRNA for the mitotic cyclins.

In summary, we identified 874 genes that show periodicexpression in the cell division cycle of a human cell line. Wefound a relationship between genes associated with prolif-eration of tumors and those that are periodically expressedduring the cell division cycle. Many of the genes that areperiodically expressed have well-characterized functions as-

sociated with the cell division cycle, and a large number,�450 clones, are either uncharacterized ESTs or hypotheticalproteins whose pattern of regulation during the cell divisioncycle now points to specific directions for further investiga-tion.

DISCUSSION

We identified 874 genes that show periodic expressionacross the human cell cycle in a well-studied cancer cell line(HeLa). This system was chosen because a high degree ofsynchrony and low background of noncycling cells can beobtained by multiple methods. Herein, 1134 clones (repre-senting 874 UNIGENE clusters) passed a minimum set ofobjective quantitative criteria for a periodically expressedgene. Although both the Rb and P53 tumor suppressors areinactivated in HeLa cells as a result of binding of the E6/E7proteins of the human papillomavirus (Scheffner et al., 1991),genes involved in basic cell cycle processes such as DNAreplication, chromosome segregation, and cell adhesion stillshowed periodic expression.

Each gene was assigned a cell cycle phase as describedabove (Figures 2 and 3). Approximately 38% of the genesidentified as cell cycle regulated were assigned a phase ofG1/S or S phase, whereas 45% were assigned a phase of G2or G2/M. In mammalian cells, G2 and M phases are short (4h of each 14–16-h cell cycle) relative to G1 and S phase, yetalmost half of the periodically expressed genes peak duringthis interval. One potential explanation is that we were moreable to detect genes expressed at G2 and M phases becausethree of our five synchronization methods arrest cells in Sphase, which provides more robust synchrony in G2 and Mphases. This possibility now seems particularly unlikely be-cause a similar distribution is observed in the proliferationclusters, which have now been observed in a variety oftumors independently.

We recognize that our microarrays do not completelyrepresent all expressed genes in the human genome, limitingthe completeness of our survey. For example, three of the 49genes known from the literature to be cell cycle-regulatedwere not represented on our arrays (see above and Table 2).Subject only to these omissions, we conclude, based on thesimilarities found between cells synchronized in S phase,cells synchronized in M phase, cells selected by mitoticshake-off, and the tumor proliferation clusters, that we haveidentified a comprehensive list of cell cycle-regulated hu-man genes.

Comparison to Previous Surveys of PeriodicallyExpressed Genes in Animal CellsGenome-wide transcriptional profiles of the cell cycle inanimal cells have been reported previously from experi-ments carried out on fibroblasts and epithelial cells. A studyof the response of primary human fibroblast to serum re-vealed not only a cell cycle response but also an equallyprominent wound-healing response (Iyer et al., 1999). Thismade clear the shortcomings of serum as a synchronizingagent if the goal is to find periodically expressed genes.Herein, we have avoided the use of serum “starvation”entirely.



A study of the cell cycle and E2F transcription factors inmouse embryo fibroblasts identified 578 cell cycle-regulatedgenes and E2F targets (Ishida et al., 2001). Many of the genesidentified in the mouse study were also identified as cellcycle-regulated in HeLa cells herein. Among these areTYMS, RRM1, RRM2, Cyclin E1, MCM3, MCM7, PCNA,TOPIIA, FEN1, RAD51, SLBP, Cyclin A2, Cyclin B1, CyclinB2, KI-67, and importin alpha 2. Those identified as E2Ftargets but not identified as cell cycle regulated in our studyinclude thymidine kinase 1 (TK), CDK2, DNA ligase I, andRB. It should be noted that not all E2F targets are expectedto be cell cycle regulated in our study of HeLa cells becausesome genes show regulation primarily during the resting togrowing transition rather than in a continuous cell cycle.

A study of the G2 DNA damage checkpoint in HeLa cellsrevealed a delay in the expression of mitotic genes when thecells are exposed to ionizing radiation (e.g., Cyclin B1, CKS2,TTK, STK15, CDC20, and CENPA) (Crawford and Piwnica-Worms, 2001). Many of these genes were classified as G2/Min our study. It is worth pointing out that similar resultswere obtained herein for G2 genes when cells are arrested inmitosis by nocodazole; G1/S, S, and G2 genes show rela-tively low expression, whereas G2/M and M/G1 genesshow relatively high expression (e.g., 0 h Thy-Noc samplesin Figures 2 and 6–8). The opposite is observed in cellsarrested in S phase with thymidine; relatively high expres-sion of G1/S and S phase genes, and relatively lower ex-pression of G2, G2/M, and M/G1 genes (Figures 2 and 6–8).The exception is the replication-dependent histone mRNAs,which, although expressed during S phase, are rapidly andcoordinately degraded when DNA synthesis is inhibited(Harris et al., 1991). Together, these observations suggestthat the gene expression program observed during a cellcycle arrest is strongly influenced by the point in the cellcycle where the arrest occurs; the genes normally expressedin that phase are usually but not always, relatively overex-pressed.

Another study of the cell cycle in primary human fibro-blasts synchronized by a double thymidine block, identified�700 cell cycle-regulated genes (Cho et al., 2001) that wehave mapped to 595 unique UNIGENE clusters. Surpris-ingly, only 96 genes were identified as cell cycle-regulated inboth studies (these comparisons are available in the supple-mental data). Of the remaining 499 genes that were identi-fied as cell cycle-regulated by Cho et al. (2001), but not in thisstudy, only 50 were not measured on our microarrays. Ofthe 778 genes identified as cell cycle regulated in this study,but not by Cho et al. (2001), only 109 were not measured ontheir microarrays. We have no ready explanation of thisdifference in results, except to note that there are differencesin the cell lineage, the microarray technology, and in theanalysis methods. As we have shown, 69 of the genes thatwere cell cycle-regulated in our study are nevertheless as-sociated with the proliferative state of tumors and cell lines,whereas only 29 of the genes reported to be cell cycle-regulated by Cho et al. (2001) appear in the tumor prolifer-ation clusters. We suspect that the most significant differ-ences may well be in the degree of cell synchrony achievedand the percentage of cells that reentered the cell cycle afterremoval of the synchronizing block. Alternatively, HeLacells, derived from a cervical carcinoma, may more closelyresemble cancers in their cell cycle program of gene expres-

sion and differ significantly from the corresponding pro-gram in a normal fibroblast.

Summing up the comparisons with previous work, webelieve that we have detected most of the genes expressedperiodically through the HeLa cell division cycle. It will beimportant to compare this list of 874 genes to similarlycomprehensive studies of more normal human cell types inthe future.

Comparison to Yeast Cell CycleComparison of the results of this study with our previousstudy of the yeast cell division cycle (Spellman et al., 1998)produced interesting results. When the “well-characterized”(i.e., those that encode curated protein sequences in RefSeq)and periodically expressed human gene sequences werecompared with the yeast genome sequence, �18% (155genes) had putative orthologs (Ball, Whitfield, and Botstein,unpublished data). Of these, only �26% (41) were clearlycell cycle regulated in yeast (Spellman et al., 1998), suggest-ing regulation at the level of transcription is not necessarilya conserved property. An example of a gene periodicallyexpressed in human cells but not in yeast is the humanCDC2 kinase (homolog of S. cerevisiae CDC28 kinase). YeastCDC28 is not regulated at the transcriptional level but ratherthe kinase activity of CDC28 is regulated (Lee and Nurse,1987). It is likely that many of the periodically expressedgenes in human cells that do not have periodically expressedcounterparts in yeast are subject to multiple layers of regu-lation (e.g., phosphorylation and proteolysis) as is alreadyknown for many well-studied cell cycle genes. Among thegenes periodically expressed in both species are many in-volved in basic processes such as DNA replication, repair,and metabolism, and mitosis. However, as noted by Spell-man et al. (1998), many of the genes that were periodicallysynthesized in yeast are not obviously involved in thesebasic processes but instead are genes involved in bud emer-gence and bud growth, which take place at particular pointsin the cell cycle because of the particular biology of buddingyeasts. Likewise, it may be that many genes are periodicallyexpressed in an animal cell for reasons that do not apply toyeast. Spellman et al. (1998) reasoned, partly from this ob-servation, that parsimony is likely to explain much of theobserved cell cycle regulation; genes are periodically ex-pressed primarily because there is special need for the geneproducts at particular points in the cell cycle. For humancells, the periodic synthesis of the cytoskeleton and celladhesion-associated genes is a particularly interesting casein point. The overall shape of an animal cell generallychanges, and tends to lose contact with its substrate, when itforms the mitotic spindle and undergoes cell division. Rea-soning in parallel to the yeast example, we suppose thatanimal cells might periodically synthesize cytoskeletal genesand adhesion factors because of the particular need tochange shape in mitosis, and subsequently reattach to sub-strate, a need not shared by the yeasts.

Cell Cycle and CancerComparison of the 874 genes that we identify as cell cycleregulated in HeLa cells to studies of human breast tumorsshows that almost 70% of the transcripts in the proliferationcluster are cell cycle regulated. We showed above that many


Vol. 13, June 2002 1995

genes involved in basic cell cycle processes are also morehighly expressed in more proliferative tumors. However, thecorrelation we found may reflect only the reality that pro-liferative tumors contain a large proportion of actively cy-cling cells.

There are nevertheless periodically expressed genes (e.g.,KRAS2, PTTG1, STK15, and PLK) (Der et al., 1982; Pei andMelmed, 1997; Smith et al., 1997; Zhou et al., 1998) that arelikely to contribute more directly to tumor phenotypes. Forexample, the two highest scoring genes in our study, STK15and Polo-like kinase (PLK), show peak expression at G2/M,are highly expressed in proliferative tumors, and have trans-forming activity in NIH3T3 cells (Holtrich et al., 1994; Smithet al., 1997; Bischoff et al., 1998; Zhou et al., 1998; Takai et al.,2001). STK15 has been shown to be amplified in humancolon cancers and in cell lines derived from many otherkinds of human tumors (Bischoff et al., 1998; Zhou et al.,1998). High expression has also been observed in the ab-sence of amplification and contributes to abnormal centro-some numbers in cell lines (Zhou et al., 1998). PLK has alsobeen implicated in centrosome maturation; injection of PLKinto either HeLa cells or human foreskin fibroblasts resultedin reduced centrosome size and abnormal chromatin distri-bution (Lane and Nigg, 1996). Amplification or overexpres-sion of STK15 and/or PLK has been postulated as a potentialcause of aneuploidy in human tumors (Lengauer et al., 1998).The high expression of genes involved in centrosome dupli-cation may contribute to the chromosomal translocationsand aneuploidy found in HeLa cells.

Although there is significant overlap between the genesfound in the proliferation cluster and those that we haveidentified as cell cycle-regulated, not all cell cycle-regulatedgenes are highly expressed in the more proliferative tumors.Some cell cycle-regulated genes are expressed in a hetero-geneous pattern with no clear relationship to proliferation inthe breast tumors studied (e.g., CDC6, CCNE1, ORC1, andBRCA1). Some cell cycle-regulated genes are expressed atlower levels in the more proliferative tumors (calponin 2,smoothelin, vinculin, and actin filament-associated protein).Many of the genes expressed at lower levels in tumors aregenes with peak expression at M/G1 and have roles incell-cell adhesion and regulation of the actin cytoskeleton.The strongly decreased expression of these genes in tumorsis often associated with increased invasion, metastasis, andultimately, poor prognosis (Rodriguez Fernandez et al., 1992;Rudiger, 1998; Engers and Gabbert, 2000). It is highly likelythat the regulation of these genes is responsive to manyother factors in addition to the cell cycle. Alternatively, therelative low level of expression of some of these genes mayactually be advantageous in the proliferative tumors.

The genes identified in this study include the moleculartargets of several different classes of chemotherapeuticagents (antimetabolite targets TYMS, RRM1 and 2, andDHFR; tubulin targeted by antimicrotubule drugs, TOP1and TOP2 inhibitors, etc.) (Ratain, 1997). The list of genesthat are strongly expressed in proliferative tumors, espe-cially those previously uncharacterized, may prove to be auseful source of additional drug targets of this kind.

In conclusion, it is important to note that we haveextracted only a fraction of the information inherent inthis dataset. The entire dataset is now freely available forany purpose (http://genome-www.stanford.edu/Human-

CellCycle/HeLa/). We are confident that others will findit a useful source for further discoveries and insights intothe cell cycle and cancer.

ACKNOWLEDGMENTS

We acknowledge members of the Botstein and Brown laboratoriesfor helpful discussions; Max Diehn, Ash Alizadeh, and JenniferBoldrick for assistance in the collection of time points; Orly Alterand Olga Troyanskaya for discussions of analysis methods; theStanford Functional Genomics Core Facility for the production ofmicroarrays; and Tim Stearns and May C. Chen for critical readingof the manuscript. M.L.W. is supported by a National ResearchService Award Postdoctoral Fellowship from the National HumanGenome Research Institute (HG00220-02) and by funds from theScleroderma Research Foundation. J.I.M. is a Howard Hughes Med-ical Institute Predoctoral Fellow. M.M.H. is supported by a grantfrom the American Cancer Society, Florida Division (F99FSU-3)with prior support from the National Institutes of Health (GM-46768). This work was supported by grants from the NationalCancer Institute to D.B. and P.O.B. (CA-77097 and CA-85129). P.O.B.is an Associate Investigator of the Howard Hughes Medical Insti-tute.

REFERENCES

Abrieu, A., Kahana, J.A., Wood, K.W., and Cleveland, D.W. (2000).CENP-E as an essential component of the mitotic checkpoint invitro. Cell 102, 817–826.

Abrieu, A., Magnaghi-Jaulin, L., Kahana, J.A., Peter, M., Castro, A.,Vigneron, S., Lorca, T., Cleveland, D.W., and Labbe, J. (2001). Mps1is a kinetochore-associated kinase essential for the vertebrate mitoticcheckpoint. Cell 106, 83–93.

Adolph, K.W., and Phelps, J.P. (1982). Role of non-histones in chro-mosome structure. Cell cycle variations in protein synthesis. J. Biol.Chem. 257, 9086–9092.

Alizadeh, A.A., et al. (2000). Distinct types of diffuse large B-celllymphoma identified by gene expression profiling. Nature 403,503–511.

Alter, O., Brown, P.O., and Botstein, D. (2000). Singular value de-composition for genome-wide expression data processing and mod-eling. Proc. Natl. Acad. Sci. USA 97, 10101–10106.

Anastasiadis, P.Z., and Reynolds, A.B. (2001). Regulation of RhoGTPases by p120-catenin. Curr. Opin. Cell Biol. 13, 604–610.

Archer, S.Y., and Hodin, R.A. (1999). Histone acetylation and can-cer. Curr. Opin. Genet. Dev. 9, 171–174.

Ashburner, M., et al. (2000). Gene ontology: tool for the unificationof biology. The Gene Ontology Consortium. Nat. Genet. 25, 25–29.

Bai, C., Richman, R., and Elledge, S.J. (1994). Human cyclin F. EMBOJ. 13, 6087–6098.

Barbacid, M. (1990). ras oncogenes: their role in neoplasia. Eur.J. Clin. Invest. 20, 225–235.

Bischoff, J.R., et al. (1998). A homologue of Drosophila aurora kinaseis oncogenic and amplified in human colorectal cancers. EMBO J. 17,3052–3065.

Bishop, J.M. (1991). Molecular themes in oncogenesis. Cell 64, 235–248.

Bjorklund, S., Skog, S., Tribukait, B., and Thelander, L. (1990). S-phase-specific expression of mammalian ribonucleotide reductaseR1 and R2 subunit mRNAs. Biochemistry 29, 5452–5458.

Carazo-Salas, R.E., Guarguaglini, G., Gruss, O.J., Segref, A.,Karsenti, E., and Mattaj, I.W. (1999). Generation of GTP-bound Ran



by RCC1 is required for chromatin-induced mitotic spindle forma-tion. Nature 400, 178–181.

Cho, R.J., et al. (1998). A genome-wide transcriptional analysis of themitotic cell cycle. Mol. Cell 2, 65–73.

Cho, R.J., Huang, M., Campbell, M.J., Dong, H., Steinmetz, L.,Sapinoso, L., Hampton, G., Elledge, S.J., Davis, R.W., and Lockhart,D.J. (2001). Transcriptional regulation and function during the hu-man cell cycle. Nat. Genet. 27, 48–54.

Cleveland, D.W. (1989). Gene regulation through messenger RNAstability. Curr. Opin. Cell Biol. 1, 1148–1153.

Crawford, D.F., and Piwnica-Worms, H. (2001). The G2 dna damagecheckpoint delays expression of genes encoding mitotic regulators.J. Biol. Chem. 276, 37166–37177.

Dalton, S. (1992). Cell cycle regulation of the human cdc2 gene.EMBO J. 11, 1797–1804.

Davenport, J.W., Fernandes, E.R., Harris, L.D., Neale, G.A., andGoorha, R. (1999). The mouse mitotic checkpoint gene Bub1b, anovel Bub1 family member, is expressed in a cell cycle-dependentmanner. Genomics 55, 113–117.

Der, C.J., Krontiris, T.G., and Cooper, G.M. (1982). Transforminggenes of human bladder and lung carcinoma cell lines are homol-ogous to the ras genes of Harvey and Kirsten sarcoma viruses. Proc.Natl. Acad. Sci. USA 79, 3637–3640.

Dominski, Z., Zheng, L.X., Sanchez, R., and Marzluff, W.F. (1999).Stem-loop binding protein facilitates 3�-end formation by stabilizingU7 snRNP binding to histone pre-mRNA. Mol. Cell Biol. 19, 3561–3570.

Eisen, M.B., and Brown, P.O. (1999). DNA arrays for analysis ofgene expression. Methods Enzymol. 303, 179–205.

Eisen, M.B., Spellman, P.T., Brown, P.O., and Botstein, D. (1998).Cluster analysis and display of genome-wide expression patterns.Proc. Natl. Acad. Sci. USA 95, 14863–14868.

Eliassen, K.A., Baldwin, A., Sikorski, E.M., and Hurt, M.M. (1998).Role for a YY1-binding element in replication-dependent mousehistone gene expression. Mol. Cell. Biol. 18, 7106–7118.

Engers, R., and Gabbert, H.E. (2000). Mechanisms of tumor metas-tasis: cell biological aspects and clinical implications. J. Cancer Res.Clin. Oncol. 126, 682–692.

Fisher, R.P., and Morgan, D.O. (1994). A novel cyclin associates withMO15/CDK7 to form the CDK-activating kinase. Cell 78, 713–724.

Fonagy, A., Swiderski, C., Wilson, A., Bolton, W., Kenyon, N., andFreeman, J.W. (1993). Cell cycle regulated expression of nucleolarantigen P120 in normal and transformed human fibroblasts. J. CellPhysiol. 154, 16–27.

Fontijn, R.D., Goud, B., Echard, A., Jollivet, F., van Marle, J.,Pannekoek, H., and Horrevoets, A.J. (2001). The human kinesin-likeprotein RB6K is under tight cell cycle control and is essential forcytokinesis. Mol. Cell Biol. 21, 2944–2955.

Georgakopoulos, A., et al. (1999). Presenilin-1 forms complexes withthe cadherin/catenin cell-cell adhesion system and is recruited tointercellular and synaptic contacts. Mol. Cell 4, 893–902.

Goswami, P.C., Roti Roti, J.L., and Hunt, C.R. (1996). The cellcycle-coupled expression of topoisomerase II� during S phase isregulated by mRNA stability and is disrupted by heat shock orionizing radiation. Mol. Cell. Biol. 16, 1500–1508.

Grosheva, I., Shtutman, M., Elbaum, M., and Bershadsky, A.D.(2001). p120 catenin affects cell motility via modulation of activity ofRho-family GTPases. A link between cell-cell contact formation andregulation of cell locomotion. J. Cell Sci. 114, 695–707.

Gruss, O.J., Carazo-Salas, R.E., Schatz, C.A., Guarguaglini, G., Kast,J., Wilm, M., Le Bot, N., Vernos, I., Karsenti, E., and Mattaj, I.W.

(2001). Ran induces spindle assembly by reversing the inhibitoryeffect of importin alpha on TPX2 activity. Cell 104, 83–93.

Guenette, S., Magendantz, M., and Solomon, F. (1995). Suppressionof a conditional mutation in alpha-tubulin by overexpression of twocheckpoint genes. J. Cell Sci. 108, 1195–1204.

Gyuris, J., Golemis, E., Chertkov, H., and Brent, R. (1993). Cdi1, ahuman G1 and S phase protein phosphatase that associates withCdk2. Cell 75, 791–803.

Harris, M.E., Bohni, R., Schneiderman, M.H., Ramamurthy, L.,Schumperli, D., Marzluff, W.F. (1991). Regulation of histone mRNAin the unperturbed cell cycle: evidence suggesting control at twoposttranscriptional steps. Mol. Cell. Biol. 11, 2416–2424.

Heintz, N., Sive, H.L., and Roeder, R.G. (1983). Regulation of humanhistone gene expression: kinetics of accumulation and changes inthe rate of synthesis and in the half-lives of individual histonemRNAs during the HeLa cell cycle. Mol. Cell. Biol. 3, 539–550.

Hetzer, M., Bilbao-Cortes, D., Walther, T.C., Gruss, O.J., and Mattaj,I.W. (2000). GTP hydrolysis by Ran is required for nuclear envelopeassembly. Mol. Cell 5, 1013–1024.

Hinchcliffe, E.H., and Sluder, G. (2001). “It takes two to tango”:understanding how centrosome duplication is regulated through-out the cell cycle. Genes Dev. 15, 1167–1181.

Hirai, H., Roussel, M.F., Kato, J.Y., Ashmun, R.A., and Sherr, C.J.(1995). Novel INK4 proteins, p19 and p18, are specific inhibitors ofthe cyclin D-dependent kinases CDK4 and CDK6. Mol. Cell. Biol. 15,2672–2681.

Hirose, K., Kawashima, T., Iwamoto, I., Nosaka, T., and Kitamura,T. (2001). MgcRacGAP is involved in cytokinesis through associat-ing with mitotic spindle and midbody. J. Biol. Chem. 276, 5821–5828.

Hiyama, H., Iavarone, A., and Reeves, S.A. (1998). Regulation of thecdk inhibitor p21 gene during cell cycle progression is under thecontrol of the transcription factor E2F. Oncogene 16, 1513–1523.

Holtrich, U., Wolf, G., Brauninger, A., Karn, T., Bohme, B., Rubsa-men-Waigmann, H., and Strebhardt, K. (1994). Induction and down-regulation of PLK, a human serine/threonine kinase expressed inproliferating cells and tumors. Proc. Natl. Acad. Sci. USA 91, 1736–1740.

Hoque, M.T., and Ishikawa, F. (2001). Human chromatid cohesincomponent hRad21 is phosphorylated in M phase and associatedwith metaphase centromeres. J. Biol. Chem. 276, 5059–5067.

Horne, M.C., Goolsby, G.L., Donaldson, K.L., Tran, D., Neubauer,M., and Wahl, A.F. (1996). Cyclin G1 and cyclin G2 comprise a newfamily of cyclins with contrasting tissue-specific and cell cycle-regulated expression. J. Biol. Chem. 271, 6050–6061.

Hunter, T. (1997). Oncoprotein networks. Cell 88, 333–346.

Ishida, S., Huang, E., Zuzan, H., Spang, R., Leone, G., West, M., andNevins, J.R. (2001). Role for E2F in control of both DNA replicationand mitotic functions as revealed from DNA microarray analysis.Mol. Cell. Biol. 21, 4684–4699.

Iyer, V.R., et al. (1999). The transcriptional program in the responseof human fibroblasts to serum. Science 283, 83–87.

Jackman, M., Firth, M., and Pines, J. (1995). Human cyclins B1 andB2 are localized to strikingly different structures: B1 to microtu-bules, B2 primarily to the Golgi apparatus. EMBO J. 14, 1646–1654.

Jinno, S., Suto, K., Nagata, A., Igarashi, M., Kanaoka, Y., Nojima, H.,and Okayama, H. (1994). Cdc25A is a novel phosphatase function-ing early in the cell cycle. EMBO J. 13, 1549–1556.

Jockusch, B.M., and Isenberg, G. (1981). Interaction of alpha-actininand vinculin with actin: opposite effects on filament network for-mation. Proc. Natl. Acad. Sci. USA 78, 3005–3009.


Vol. 13, June 2002 1997

Kaczmarek, L. (1986). Protooncogene expression during the cellcycle. Lab. Invest. 54, 365–376.

Kimura, M., Kotani, S., Hattori, T., Sumi, N., Yoshioka, T.,Todokoro, K., and Okano, Y. (1997). Cell cycle-dependent expres-sion and spindle pole localization of a novel human protein kinase,Aik, related to Aurora of Drosophila and yeast Ipl1. J. Biol. Chem.272, 13766–13771.

Kinch, M.S., Clark, G.J., Der, C.J., and Burridge, K. (1995). Tyrosinephosphorylation regulates the adhesions of ras-transformed breastepithelia. J. Cell Biol. 130, 461–471.

Knehr, M., Poppe, M., Enulescu, M., Eickelbaum, W., Stoehr, M.,Schroeter, D., and Paweletz, N. (1995). A critical appraisal of syn-chronization methods applied to achieve maximal enrichment ofHeLa cells in specific cell cycle phases. Exp. Cell Res. 217, 546–553.

Kondili, K., Tsolas, O., and Papamarcaki, T. (1996). Selective inter-action between parathymosin and histone H1. Eur. J. Biochem. 242,67–74.

Lane, H.A., and Nigg, E.A. (1996). Antibody microinjection revealsan essential role for human polo-like kinase 1 (Plk1) in the func-tional maturation of mitotic centrosomes. J. Cell Biol. 135, 1701–1713.

Laub, M.T., McAdams, H.H., Feldblyum, T., Fraser, C.M., and Sha-piro, L. (2000). Global analysis of the genetic network controlling abacterial cell cycle. Science 290, 2144–2148.

Lee, M.G., and Nurse, P. (1987). Complementation used to clone ahuman homologue of the fission yeast cell cycle control gene cdc2.Nature 327, 31–35.

Lee, K.S., Yuan, Y.L., Kuriyama, R., and Erikson, R.L. (1995). Plk isan M-phase-specific protein kinase and interacts with a kinesin-likeprotein, CHO1/MKLP-1. Mol. Cell. Biol. 15, 7143–7151.

Lei, M., and Tye, B.K. (2001). Initiating DNA synthesis. from recruit-ing to activating the MCM complex. J. Cell Sci. 114, 1447–1454.

Lengauer, C., Kinzler, K.W., and Vogelstein, B. (1998). Genetic in-stabilities in human cancers. Nature 396, 643–649.

Leone, G., DeGregori, J., Yan, Z., Jakoi, L., Ishida, S., Williams, R.S.,and Nevins, J.R. (1998). E2F3 activity is regulated during the cellcycle and is required for the induction of S phase. Genes Dev. 12,2120–2130.

Lew, D.J., Dulic, V., and Reed, S.I. (1991). Isolation of three novelhuman cyclins by rescue of G1 cyclin (Cln) function in yeast. Cell 66,1197–1206.

Li, F., and Altieri, D.C. (1999). The cancer antiapoptosis mousesurvivin gene: characterization of locus and transcriptional require-ments of basal and cell cycle-dependent expression. Cancer Res. 59,3143–3151.

Li, J., Xu, M., Zhou, H., Ma, J., and Potter, H. (1997). Alzheimerpresenilins in the nuclear membrane, interphase kinetochores, andcentrosomes suggest a role in chromosome segregation. Cell 90,917–927.

Liao, H., Winkfein, R.J., Mack, G., Rattner, J.B., and Yen, T.J. (1995).CENP-F is a protein of the nuclear matrix that assembles ontokinetochores at late G2 and is rapidly degraded after mitosis. J. CellBiol. 130, 507–518.

Lucibello, F.C., Truss, M., Zwicker, J., Ehlert, F., Beato, M., andMuller, R. (1995). Periodic Cdc25C transcription is mediated by anovel cell cycle-regulated repressor element (CDE). EMBO J. 14,132–142.

Ma, T., et al. (2000). Cell cycle-regulated phosphorylation ofp220(NPAT) by cyclin E/Cdk2 in Cajal bodies promotes histonegene transcription. Genes Dev. 14, 2298–2313.

Marra, G., Chang, C.L., Laghi, L.A., Chauhan, D.P., Young, D., andBoland, C.R. (1996). Expression of human MutS homolog 2 (hMSH2)protein in resting and proliferating cells. Oncogene 13, 2189–2196.

Marzluff, W.F. (1992). Histone 3� ends: essential and regulatoryfunctions. Gene Expression 2, 93–97.

Marzluff, W.F., and Pandey, N.B. (1988). Multiple levels of regula-tion of histone mRNA concentrations. Trends Biochem. Sci. 13,49–52.

Masuda, H., Tanaka, K., Takagi, M., Ohgami, K., Sakamaki, T.,Shibata, N., and Takahashi, K. (1996). Molecular cloning and char-acterization of human non-smooth muscle calponin. J. Biochem. 120,415–424.

Matsuoka, M., Kato, J.Y., Fisher, R.P., Morgan, D.O., and Sherr, C.J.(1994). Activation of cyclin-dependent kinase 4 (cdk4) by mouseMO15-associated kinase. Mol. Cell. Biol. 14, 7265–7275.

Mitchell, J.R., Wood, E., and Collins, K. (1999). A telomerase com-ponent is defective in the human disease dyskeratosis congenita.Nature 402, 551–555.

Morris, G.F., and Mathews, M.B. (1989). Regulation of proliferatingcell nuclear antigen during the cell cycle. J. Biol. Chem. 264, 13856–13864.

Muchardt, C., and Yaniv, M. (2001). When the SWI/SNF complexremodels . . . the cell cycle. Oncogene 20, 3067–3075.

Nachury, M.V., Maresca, T.J., Salmon, W.C., Waterman-Storer,C.M., Heald, R., and Weis, K. (2001). Importin beta is a mitotic targetof the small GTPase Ran in spindle assembly. Cell 104, 95–106.

Nurse, P. (2000). A long twentieth century of the cell cycle andbeyond. Cell 100, 71–78.

Ohtani, K., DeGregori, J., and Nevins, J.R. (1995). Regulation of thecyclin E gene by transcription factor E2F1. Proc. Natl. Acad. Sci.USA 92, 12146–12150.

Okuda, T., Hirai, H., Valentine, V.A., Shurtleff, S.A., Kidd, V.J.,Lahti, J.M., Sherr, C.J., and Downing, J.R. (1995). Molecular cloning,expression pattern, and chromosomal localization of humanCDKN2D/INK4d, an inhibitor of cyclin D-dependent kinases.Genomics 29, 623–630.

Ouyang, B., Lan, Z., Meadows, J., Pan, H., Fukasawa, K., Li, W., andDai, W. (1998). Human Bub1: a putative spindle checkpoint kinaseclosely linked to cell proliferation. Cell Growth Differ. 9, 877–885.

Pei, L., and Melmed, S. (1997). Isolation and characterization of apituitary tumor-transforming gene (PTTG). Mol. Endocrinol. 11,433–441.

Perou, C.M., et al. (1999). Distinctive gene expression patterns inhuman mammary epithelial cells and breast cancers. Proc. Natl.Acad. Sci. USA 96, 9212–9217.

Perou, C.M., et al. (2000). Molecular portraits of human breast tu-mors. Nature 406, 747–752.

Pines, J., and Hunter, T. (1989). Isolation of a human cyclin cDNA:evidence for cyclin mRNA and protein regulation in the cell cycleand for interaction with p34cdc2. Cell 58, 833–846.

Pines, J., and Hunter, T. (1990). Human cyclin A is adenovirusE1A-associated protein p60 and behaves differently from cyclin B.Nature 346, 760–763.

Plumb, M., Stein, J., and Stein, G. (1983). Coordinate regulation ofmultiple histone mRNAs during the cell cycle in HeLa cells. NucleicAcids Res. 11, 2391–2410.

Pruitt, K.D., and Maglott, D.R. (2001). RefSeq, and LocusLink. NCBIgene-centered resources. Nucleic Acids Res. 29, 137–140.

Puck, T.T., Marcus, P.I., and Cieciura, S.J. (1956). Clonal growth ofmammalian cells in vitro. J. Exp. Med. 103, 273–284.



Rajan, J.V., Wang, M., Marquis, S.T., and Chodosh, L.A. (1996).Brca2 is coordinately regulated with Brca1 during proliferation anddifferentiation in mammary epithelial cells. Proc. Natl. Acad. Sci.USA 93, 13078–13083.

Ratain, M.J. (1997). Pharmacology of cancer chemotherapy. In: Can-cer: Principles and Practice of Oncology, ed. V.T. DeVita, S. Hell-man, S.A. Rosenberg, Philadelphia: Lippincott-Raven Publishers,375–512.

Richardson, R.T., Batova, I.N., Widgren, E.E., Zheng, L.X., Whit-field, M., Marzluff, W.F., and O’Rand, M.G. (2000). Characterizationof the histone H1-binding protein, NASP, as a cell cycle-regulatedsomatic protein. J. Biol. Chem. 275, 30378–30386.

Richardson, H.E., Stueland, C.S., Thomas, J., Russell, P., and Reed,S.I. (1990). Human cDNAs encoding homologs of the smallp34Cdc28/Cdc2-associated protein of Saccharomyces cerevisiae andSchizosaccharomyces pombe. Genes Dev. 4, 1332–1344.

Rodriguez Fernandez, J.L., Geiger, B., Salomon, D., Sabanay, I.,Zoller, M., and Ben Ze’ev, A. (1992). Suppression of tumorigenicityin transformed cells after transfection with vinculin cDNA. J. CellBiol. 119, 427–438.

Ross, D.T., et al. (2000). Systematic variation in gene expressionpatterns in human cancer cell lines. Nat. Genet. 24, 227–235.

Rudiger, M. (1998). Vinculin and �-catenin: shared and uniquefunctions in adherens junctions. Bioessays 20, 733–740.

Saha, P., Thome, K.C., Yamaguchi, R., Hou, Z., Weremowicz, S., andDutta, A. (1998). The human homolog of Saccharomyces cerevisiaeCDC45. J. Biol. Chem. 273, 18205–18209.

Sardet, C., Vidal, M., Cobrinik, D., Geng, Y., Onufryk, C., Chen, A.,and Weinberg, R.A. (1995). E2F-4 and E2F-5, two members of theE2F family, are expressed in the early phases of the cell cycle. Proc.Natl. Acad. Sci. USA 92, 2403–2407.

Scheffner, M., Munger, K., Byrne, J.C., and Howley, P.M. (1991). Thestate of the p53 and retinoblastoma genes in human cervical carci-noma cell lines. Proc. Natl. Acad. Sci. USA 88, 5523–5527.

Schneiderman, M.H., Dewey, W.C., Leeper, D.B., and Nagasawa, H.(1972). Use of the mitotic selection procedure for cell cycle analysis.Comparison between the X-ray and cycloheximide G2 markers.Exp. Cell Res. 74, 430–438.

Schumperli, D. (1988). Multilevel regulation of replication-depen-dent histone genes. Trends Genet. 4, 187–191.

Shah, J.V., and Cleveland, D.W. (2000). Waiting for anaphase: Mad2and the spindle assembly checkpoint. Cell 103, 997–1000.

Shelby, R.D., Vafa, O., and Sullivan, K.F. (1997). Assembly ofCENP-A into centromeric chromatin requires a cooperative array ofnucleosomal DNA contact sites. J. Cell Biol. 136, 501–513.

Sherlock, G., et al. (2001). The Stanford Microarray Database. Nu-cleic Acids Res. 29, 152–155.

Shindo, M., Nakano, H., Kuroyanagi, H., Shirasawa, T., Mihara, M.,Gilbert, D.J., Jenkins, N.A., Copeland, N.G., Yagita, H., Okumura, K.(1998). cDNA cloning, expression, subcellular localization, andchromosomal assignment of mammalian aurora homologues, auro-ra-related kinase (ARK) 1 and 2. Biochem. Biophys. Res. Commun.244, 285–292.

Slansky, J.E., and Farnham, P.J. (1996). Transcriptional regulation ofthe dihydrofolate reductase gene. Bioessays 18, 55–62.

Smith, M.R., Wilson, M.L., Hamanaka, R., Chase, D., Kung, H.,Longo, D.L., and Ferris, D.K. (1997). Malignant transformation ofmammalian cells initiated by constitutive expression of the polo-likekinase. Biochem. Biophys. Res. Commun. 234, 397–405.

Spellman, P.T., Sherlock, G., Zhang, M.Q., Iyer, V.R., Anders, K.,Eisen, M.B., Brown, P.O., Botstein, D., and Futcher, B. (1998). Com-

prehensive identification of cell cycle-regulated genes of the yeastSaccharomyces cerevisiae by microarray hybridization. Mol. Biol. Cell9, 3273–3297.

Stampfer, M.R., Pan, C.H., Hosoda, J., Bartholomew, J., Mendel-sohn, J., Yaswen, P. (1993). Blockage of EGF receptor signal trans-duction causes reversible arrest of normal and immortal humanmammary epithelial cells with synchronous reentry into the cellcycle. Exp. Cell Res. 208, 175–188.

Stillman, B. (1996). Cell cycle control of DNA replication. Science274, 1659–1664.

Takai, N., Miyazaki, T., Fujisawa, K., Nasu, K., Hamanaka, R., andMiyakawa, I. (2001). Expression of polo-like kinase in ovarian can-cer is associated with histological grade and clinical stage. CancerLett. 164, 41–49.

Tassan, J.P., Schultz, S.J., Bartek, J., and Nigg, E.A. (1994). Cell cycleanalysis of the activity, subcellular localization, and subunit com-position of human CAK (CDK-activating kinase). J. Cell Biol. 127,467–478.

Tobey, R.A., Valdez, J.G., and Crissman, H.A. (1988). Synchroniza-tion of human diploid fibroblasts at multiple stages of the cell cycle.Exp. Cell Res. 179, 400–416.

Troyanskaya, O., Cantor, M., Sherlock, G., Brown, P., Hastie, T.,Tibshirani, R., Botstein, D., and Altman, R.B. (2001). Missing valueestimation methods for DNA microarrays. Bioinformatics 17, 520–525.

Vareli, K., Frangou-Lazaridis, M., van, d.K., I, Tsolas, O., and vanDriel, R. (2000). Nuclear distribution of prothymosin alpha andparathymosin: evidence that prothymosin alpha is associated withRNA synthesis processing and parathymosin with early DNA rep-lication. Exp. Cell Res. 257, 152–161.

Vaughn, J.P., Cirisano, F.D., Huper, G., Berchuck, A., Futreal, P.A.,Marks, J.R., and Iglehart, J.D. (1996). Cell cycle control of BRCA2.Cancer Res. 56, 4590–4594.

Wang, E., Miller, L.D., Ohnmacht, G.A., Liu, E.T., and Marincola,F.M. (2000). High-fidelity mRNA amplification for gene profiling.Nat. Biotechnol. 18, 457–459.

Wang, Z.-F., Whitfield, M.L., Ingledue, T.I., Dominski, Z., andMarzluff, W.F. (1996a). The protein which binds the 3� end ofhistone mRNA: a novel RNA-binding protein required for histonepre-mRNA processing. Genes Dev. 10, 3028–3040.

Wang, W., Xue, Y., Zhou, S., Kuo, A., Cairns, B.R., and Crabtree,G.R. (1996b). Diversity and specialization of mammalian SWI/SNFcomplexes. Genes Dev. 10, 2117–2130.

Weinstein, J. (1997). Cell cycle-regulated expression, phosphoryla-tion, and degradation of p55Cdc. A mammalian homolog ofCDC20/Fizzy/slp1. J. Biol. Chem. 272, 28501–28511.

Weinstein, J., Jacobsen, F.W., Hsu-Chen, J., Wu, T., and Baum, L.G.(1994). A novel mammalian protein, p55CDC, present in dividingcells is associated with protein kinase activity and has homology tothe Saccharomyces cerevisiae cell division cycle proteins Cdc20 andCdc4. Mol. Cell. Biol. 14, 3350–3363.

Wells, D., and Kedes, L. (1985). Structure of a human histone cDNA:evidence that basally expressed histone genes have interveningsequences and encode polyadenylylated mRNAs. Proc. Natl. Acad.Sci. USA 82, 2834–2838.

Whitfield, M.L., Zheng, L.X., Baldwin, A., Ohta, T., Hurt, M.M., andMarzluff, W.F. (2000). Stem-loop binding protein, the protein thatbinds the 3� end of histone mRNA, is cell cycle regulated by bothtranslational and posttranslational mechanisms. Mol. Cell. Biol. 20,4188–4198.


Vol. 13, June 2002 1999

Wiese, C., Wilde, A., Moore, M.S., Adam, S.A., Merdes, A., andZheng, Y. (2001). Role of importin-beta in coupling Ran to down-stream targets in microtubule assembly. Science 291, 653–656.

Wilkins, J.A., and Lin, S. (1982). High-affinity interaction of vinculinwith actin filaments in vitro. Cell 28, 83–90.

Williams, G.H., Romanowski, P., Morris, L., Madine, M., Mills, A.D.,Stoeber, K., Marr, J., Laskey, R.A., and Coleman, N. (1998). Im-proved cervical smear assessment using antibodies against proteinsthat regulate DNA replication. Proc. Natl. Acad. Sci. USA 95, 14932–14937.

Wittmann, T., Hyman, A., and Desai, A. (2001). The spindle: adynamic assembly of microtubules and motors. Nat. Cell Biol. 3,E28–E34.

Wood, K.W., Sakowicz, R., Goldstein, L.S., and Cleveland, D.W.(1997). CENP-E is a plus end-directed kinetochore motor requiredfor metaphase chromosome alignment. Cell 91, 357–366.

Yamamoto, T., Harada, N., Kano, K., Taya, S., Canaani, E., Mat-suura, Y., Mizoguchi, A., Ide, C., and Kaibuchi, K. (1997). The Rastarget AF-6 interacts with ZO-1 and serves as a peripheral compo-nent of tight junctions in epithelial cells. J. Cell Biol. 139, 785–795.

Yamamoto, T., Taya, S., and Kaibuchi, K. (1999). Ras-induced trans-formation and signaling pathway. J. Biochem. 126, 799–803.

Yan, Z., DeGregori, J., Shohet, R., Leone, G., Stillman, B., Nevins,J.R., and Williams, R.S. (1998). Cdc6 is regulated by E2F and isessential for DNA replication in mammalian cells. Proc. Natl. Acad.Sci. USA 95, 3603–3608.

Zariwala, M., Liu, J., and Xiong, Y. (1998). Cyclin E2, a novel humanG1 cyclin and activating partner of CDK2 and CDK3, is induced byviral oncoproteins. Oncogene 17, 2787–2798.Zhao, J., Dynlacht, B., Imai, T., Hori, T., and Harlow, E. (1998).Expression of NPAT, a novel substrate of cyclin E-CDK2, promotesS-phase entry. Genes Dev. 12, 456–461.Zhao, J., Kennedy, B.K., Lawrence, B.D., Barbie, D.A., Matera, A.G.,Fletcher, J.A., and Harlow, E. (2000). NPAT links cyclin E-Cdk2 tothe regulation of replication-dependent histone gene transcription.Genes Dev. 14, 2283–2297.Zhong, C., Kinch, M.S., and Burridge, K. (1997). Rho-stimulatedcontractility contributes to the fibroblastic phenotype of Ras-trans-formed epithelial cells. Mol. Biol. Cell 8, 2329–2344.Zhou, H., Kuang, J., Zhong, L., Kuo, W.L., Gray, J.W., Sahin, A.,Brinkley, B.R., and Sen, S. (1998). Tumor amplified kinase STK15/BTAK induces centrosome amplification, aneuploidy and transfor-mation. Nat. Genet. 20, 189–193.Zieve, G.W., Turnbull, D., Mullins, J.M., and McIntosh, J.R. (1980).Production of large numbers of mitotic mammalian cells by use ofthe reversible microtubule inhibitor nocodazole. Nocodazole accu-mulated mitotic cells. Exp. Cell Res. 126, 397–405.Zou, H., McGarry, T.J., Bernal, T., and Kirschner, M.W. (1999).Identification of a vertebrate sister-chromatid separation inhibitorinvolved in transformation and tumorigenesis. Science 285, 418–422.Zwicker, J., Lucibello, F.C., Wolfraim, L.A., Gross, C., Truss, M.,Engeland, K., and Muller, R. (1995). Cell cycle regulation of thecyclin A, cdc25C and cdc2 genes is based on a common mechanismof transcriptional repression. EMBO J. 14, 4514–4522.



Documents

Identification of Genes Periodically Expressed in the Human Cell